IMAGE

ENCRYPTION
A Communication Perspective

Fathi E. Abd El-Samie u Hossam Eldin H. Ahmed

Ibrahim F. Elashry u Mai H. Shahieen
Osama S. Faragallah u El-Sayed M. El-Rabaie
Saleh A. Alshebeili

www.allitebooks.com
 IMAGE
ENCRYPTION
A Communication Perspective

www.allitebooks.com
www.allitebooks.com
 IMAGE
ENCRYPTION
A Communication Perspective

Fathi E. Abd El-Samie u Hossam Eldin H. Ahmed

Ibrahim F. Elashry u Mai H. Shahieen
Osama S. Faragallah u El-Sayed M. El-Rabaie
Saleh A. Alshebeili

www.allitebooks.com
MATLAB® is a trademark of The MathWorks, Inc. and is used with permission. The MathWorks does not
warrant the accuracy of the text or exercises in this book. This book’s use or discussion of MATLAB® soft-
ware or related products does not constitute endorsement or sponsorship by The MathWorks of a particular
pedagogical approach or particular use of the MATLAB® software.

CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742

© 2014 by Taylor & Francis Group, LLC

CRC Press is an imprint of Taylor & Francis Group, an Informa business

No claim to original U.S. Government works

Version Date: 20131009

International Standard Book Number-13: 978-1-4665-7699-5 (eBook - PDF)

This book contains information obtained from authentic and highly regarded sources. Reasonable efforts
have been made to publish reliable data and information, but the author and publisher cannot assume
responsibility for the validity of all materials or the consequences of their use. The authors and publishers
have attempted to trace the copyright holders of all material reproduced in this publication and apologize to
copyright holders if permission to publish in this form has not been obtained. If any copyright material has
not been acknowledged please write and let us know so we may rectify in any future reprint.

Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmit-
ted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented,
including photocopying, microfilming, and recording, or in any information storage or retrieval system,
without written permission from the publishers.

For permission to photocopy or use material electronically from this work, please access www.copyright.
com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood
Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that provides licenses and
registration for a variety of users. For organizations that have been granted a photocopy license by the CCC,
a separate system of payment has been arranged.

Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used
only for identification and explanation without intent to infringe.
Visit the Taylor & Francis Web site at
http://www.taylorandfrancis.com

and the CRC Press Web site at

http://www.crcpress.com

www.allitebooks.com
Contents

P r e fa c e ix
About the A u t h o r s xi

C h a p t e r 1 I n t r o d u c t i o n 1

C h a p t e r 2 F u n d a m e n ta l s o f I m a g e E n c r y p t i o n 3
2.1 Introduction 3
2.2 Basic Concepts of Cryptography 4
2.2.1  Goals of Cryptography 4
2.2.2 Principles of Encryption 5
2.3 Classification of Encryption Algorithms 6
2.3.1 Classification According to Encryption Structure 6
2.3.2 Classification According to Keys 7
2.3.3 Classification According to Percentage of
Encrypted Data 8
2.4 Cryptanalysis 9
2.5 Features of Image Encryption Schemes 10
2.6 Conventional Symmetric Block Ciphers 11
2.6.1 Data Encryption Standard 11
2.6.2 Double DES 14
2.6.3 Triple DES 14
2.6.4 International Data Encryption Algorithm 15
2.6.5 Blowfish 16
2.6.6 RC5 Algorithm 16
2.6.6.1 RC5 Encryption Algorithm 17
2.6.6.2 RC5 Decryption Algorithm 17
2.6.6.3 RC5 Key Expansion 17
2.6.7 RC6 Algorithm 19

www.allitebooks.com
vi C o n t en t s

2.6.7.1 RC6 Encryption Algorithm 19

2.6.7.2 RC6 Decryption Algorithm 20
2.6.8 The Advanced Encryption Standard 21
2.7 Modes of Operation 24
2.7.1 The ECB Mode 25
2.7.2 The CBC Mode 26
2.7.3 The CFB Mode 27
2.7.4 The OFB Mode 27
2.8 Chaos and Cryptography 28
2.9 The 2D Chaotic Baker Map 31

C h a p t e r 3 E n c r y p t i o n E va l uat i o n M e t r i c s 33
3.1 Introduction 33
3.2 Encryption Evaluation Metrics 34
3.2.1 Histogram Deviation 34
3.2.2 Correlation Coefficient 35
3.2.3 Irregular Deviation 35
3.2.4 Histogram Uniformity 36
3.2.5 Deviation from Ideality 36
3.2.6 Avalanche Effect 37
3.2.7 NPCR and UACI 37
3.3 Other Tests 38
3.3.1 Noise Immunity 38
3.3.2 The Processing Time 38
3.4 Testing the Evaluation Metrics 38
3.5 Summary 42

C h a p t e r 4 H o m o m o r p h i c I m a g e E n c r y p t i o n 43
4.1 Overview 43
4.2 Homomorphic Cryptosystem 43
4.3 Security Analysis and Test Results 45
4.3.1 Statistical Analysis 45
4.3.1.1 Histograms of Encrypted Images 46
4.3.1.2 Encryption Quality Measurements 47
4.3.2 Key Space Analysis 48
4.3.2.1 Exhaustive Key Search 48
4.3.2.2 Key Sensitivity Test 49
4.3.3 Differential Analysis 53
4.4 Effect of Noise 54
4.5 Summary 54

C h a p t e r 5 C h a o t i c I m a g e E n c r y p t i o n w i t h D i f f e r e n t
M o d e s o f O p e r at i o n 57
5.1 Overview 57
5.2 Chaotic Encryption with Modes of Operation 57
5.3 Implementation Issues 58
5.4 Simulation Examples and Discussion 60
5.5 Summary 68

www.allitebooks.com
 C o n t en t s vii

C h a p t e r 6 D i f f u s i o n M e c h a n i s m f o r D ata E n c r y p t i o n
i n t h e ECB M o d e 71
6.1 Introduction 71
6.2 The Preprocessing Network 72
6.2.1 The Addition Part 72
6.2.2 The SPN 73
6.3 Implementation Issues 73
6.3.1 Effect of the Normalized Block Size on Diffusion 74
6.3.2 Effect of the Normalized Block Size on Block
Dependency 77
6.4 Simulation Examples 78
6.4.1 Encryption Quality 79
6.4.2 Diffusion 79
6.4.3 Encryption of Images with Few Details 82
6.5 Summary 88

C h a p t e r 7 O r t h o g o n a l F r e q u e n cy D i v i s i o n
M u lt ip l e x i n g 89
7.1 Introduction 89
7.2 Basic Principles of OFDM 90
7.2.1 Orthogonality 92
7.2.2 Frequency Domain Orthogonality 93
7.3 OFDM System Model 94
7.3.1 FFT-OFDM 94
7.3.2 DCT-OFDM 95
7.3.3 DWT–OFDM 95
7.3.4 Discrete Wavelet Transform 96
7.3.4.1 Implementation of the DWT 96
7.3.4.2 Haar Wavelet Transform 98
7.4 Guard Interval Insertion 99
7.5 Communication Channels 100
7.5.1 Additive White Gaussian Noise Channel 100
7.5.2 Fading Channel 100
7.6 Channel Estimation and Equalization 101

C h a p t e r 8 OFDM L i m i tat i o n s 103

8.1 Introduction 103
8.2 Analysis of Intercarrier Interference 104
8.3 CFO in FFT-OFDM System Model 105
8.4 CFO in DCT-OFDM System Model 107
8.5 CFO in DWT-OFDM System Model 108
8.6 CFO Compensation 108
8.7 Simulation Parameters 109
8.8 Effect of PAPR 114
8.9 PAPR Reduction Methods 119
8.9.1 The Clipping Method 120
8.9.2 The Companding Method 121
8.9.3 The Hybrid Clipping-and-Companding Method 124

www.allitebooks.com
viii C o n t en t s

8.10 Simulation Experiments of PAPR Reduction Methods 124

8.11 Sampling Clock Offset 128
8.11.1 System Model 129

C h a p t e r 9 S i m u l at i o n E x a m p l e s 133
9.1 Simulation Parameters 133
9.2 Simulation Experiments in the Presence of CFO 169
9.3 Simulation Experiments for Enhanced Algorithms 171
9.4 Simulation Experiments of PAPR Reduction Methods 187
9.5 Summary 223

R e f e r e n c e s 225
A pp e n d i x A 237
A pp e n d i x B 293

www.allitebooks.com
Preface

We try in this book to look at image encryption with the eyes of

communication researchers. Traditional studies of encryption
­
­concentrate on the strength of the encryption algorithm without
taking into consideration what is after encryption. What is after
encryption is the question we must answer to select the appropriate
encryption algorithm. For real-life applications, what is after encryp-
tion is communication of encrypted images. With the advances in
mobile and TV applications, we have to transmit encrypted images
wirelessly. So, “Do our encryption algorithms tolerate the wireless
communication impairments?” is the question we are trying to answer
in this book.
We can summarize the main contributions in this book as follows:
This book is devoted to the issue of image encryption for the
purpose of wireless communications.
Diffusion as well as permutation ciphers are considered in
this book, with a comparison between them using different
­evaluation metrics.
Modifications are presented to existing block ciphers either to
speed up or to enhance their performance.
The wireless communication environment in which the encrypted
images needs to be communicated is studied.

ix

www.allitebooks.com
x P refac e

Simulation experiments are presented for the validation of the

discussed algorithms and their modifications and for investi-
gating the performance of algorithms over wireless channels.
MATLAB® codes for most of the simulation experiments in this
book are included in two appendices at the end of the book.
Finally, we hope that this book will be helpful for the image-
processing and wireless communication communities.
MATLAB® and Simulink® are registered rademarks of The Math
Works, Inc. For product information, please contact:

The Math Works, Inc.

3 Apple Hill Drive
Natick, MA 01760-2098 USA
Tel: 508 647 7000
Fax: 508-647-7001
E-mail: [email protected]
Web: http://www.mathworks.com/
About the Authors

Fathi E. Abd El-Samie received the

BSc(Honors), MSc, and PhD degrees from
Menoufia University, Menouf, Egypt in
1998, 2001, and 2005, respectively. Since
2005, he has been a teaching staff mem-
ber with the Department of Electronics
and Electrical Communications, Faculty
of Electronic Engineering, Menoufia
University. He is currently a researcher
at KACST-TIC in radio frequency and
­photonics for the e-Society­(RFTONICs). He is a co-author of
about 200 papers in international conference proceedings and
journals and 4 textbooks. His current research interests include
image ­ enhancement, image ­ restoration, image interpolation,
­superresolution reconstruction of images, data hiding, multimedia­
communications, medical image processing, optical signal
­
processing,­and digital communications.
Dr. Abd El-Samie was a recipient of the Most Cited Paper Award
from the Digital Signal Processing journal in 2008.

Hossam Eldin H. Ahmed received a BSC(Honors) in nuclear

engineering in June 1969 (Faculty of Engineering, Alexandria

xi
x ii A b o u t t he Au t h o rs

University, Egypt); an MSc in

microelectronic ­
­ electron diffraction in
April 1977 (Nuclear Department, Faculty
of Engineering, Alexandria University);
and a PhD in June 1983 (High Institute
of Electronic and Optics, Paul Sabatier
University, Toulouse, France). From 1970
to 1977, he was in the Egyptian marine
forces. He was a demonstrator in 1977,
giving­lectures to staff members in the Department of Electronic
and Electrical Communications, Faculty of Engineering and
Technology, Menoufia University. In 1993, he became a professor
of microelectronics, VLSI design technology, communication sys-
tems, and computer networks. From 1993 until 1999, he was vice
dean for education and student affairs at the Faculty of Electronic
Engineering. In 2001, he became a chairman of the Electronics
and Electrical Communications Department. He is a member of
the Menoufia periodic electronic faculty journal and since 1995
has been the director, designer, and constructor of the Menoufia
University wide-area network (WAN) (21 LANs). He is the devel-
oper of the Menoufia University libraries and FRCU universities
libraries. His current research interests are e­lectron microscopy;
­transmission and backscattering of electrons and ion beams into
amorphous or polycrystalline targets; optical fibers; VLSI design;
nanotechnology; lithography; digital, o­ ptical, and multimedia com-
munications; digital images; multimedia and ­database communica-
tions; security applications; telemetry m­ icrocomputer applications
in satellites; and OBC and satellite communications.

Ibrahim F. Elashry graduated from the

Faculty of Engineering, Kafrelshiekh
University, Egypt in 2007. He is now a
teaching assistant and PhD student at
the University of Wollongong (UOW),
Australia. His research interests are ­security
over wired and wireless networks and image
processing.
 A b o u t t he Au t h o rs x iii

Mai H. Shahieen graduated in May 2005

from the Faculty of Electronic Engineering,
Menoufia University. Her 2011 MSc degree
is in encrypted image transmission over wire-
less channels. She is now a PhD student at
the Faculty of Electronic Engineering. Her
research interests are b ­roadband wireless
distribution systems, image and video com-
pression, multimedia systems, and wireless
networks.

Osama S. Faragallah received BSc, MSc,

and PhD degrees in computer s­cience and
engineering from Menoufia University,
­
Menouf, Egypt in 1997, 2002, and 2007,
respectively. He is currently associate p
­ rofessor
with the Department of Computer Science
and Engineering, Faculty of Electronic
Engineering, Menoufia University, where he
was a demonstrator from 1997 to 2002 and
assistant lecturer from 2002 to 2007. Since
2007, he has been a teaching staff member
with the same department. His research interests are network secu-
rity, cryptography, Internet security, multimedia security, image
encryption, watermarking, steganography, data hiding, medical
image ­processing, and chaos theory.

El-Sayed M. El-Rabaie (senior m ­ ember,

IEEE 1992, MIEE chartered electrical
­engineer) was born in Sires Elian (Menoufia),
Egypt in 1953. He received the PhD degree
in microwave device engineering from the
Queen’s University of Belfast in 1986. He was
a postdoctoral fellow at Queen’s (Department
of Electronic Engineering) until February
1989. In his doctoral research, he constructed
a CAD (computer-aided design) pack-
age in nonlinear circuit simulations based
xiv A b o u t t he Au t h o rs

on harmonic balance techniques. Since then, he has been involved

in different research areas, including CAD of nonlinear microwave
circuits, nanotechnology, digital communication systems, and digital
image processing. He was invited in 1992 as a research fellow at North
Arizona University (College of Engineering and Technology) and
in 1994 as a visiting professor at Ecole Polytechnique of Montreal,
Quebec, Canada. Professor El-Rabaie has authored and co-authored
more than 130 papers and technical reports and 15 books. In 1993,
he was awarded the Egyptian Academic Scientific Research Award
(Salah Amer Award of Electronics), and, in 1995, he received the award
of the Best Researcher on CAD from Menoufia University. He acts
as a reviewer and member of the editorial board for several ­scientific
journals. He participated in translating the first part of the Arabic
encyclopedia. Professor El-Rabaie was the head of the Electronics
and Electrical Communications Engineering Department, Faculty
of Electronic Engineering, Menoufia University, and then the vice
dean of postgraduate studies and research in the same faculty. He
currently is the vice dean of the Scientific Committee for Professors
and Assistant Professors promotion in Egypt.

Saleh A. Alshebeili is professor and

chairman (2001–2005) of the Electrical
Engineering Department, King Saud
University, Riyadh, Saudi Arabia. He has more
than 20 years of teaching and research experi-
ence in the area of communications and signal
processing. Dr. Alshebeili is a member of the
board of directors of Prince Sultan Advanced
Technologies Research Institute (PSATRI) and has been the vice
president of PSATRI (2008–2011), the director of Saudi-Telecom
­
Research Chair (2008–2012), and the director (2011–present) of the
Technology Innovation Center, RF and Photonics in the e-Society­
(RFTONICS), funded by King Abdulaziz City for Science and
Technology (KACST). Dr. Alshebeili has been on the editorial board of
the Journal of Engineering Sciences of King Saud University (­ 2009–2012).
He also has active involvement in the review process of a number of
research journals, KACST general ­directorate grants ­programs, and
national and international symposiums and conferences.
 1
I ntroducti on

The presence of communication networks has prompted new ­problems

with security and privacy. Having a secure and reliable means for
communicating with images and video is becoming a necessity, and
its related issues must be carefully considered. Hence, network secu-
rity and data encryption have become important. Images can now be
considered one of the most usable forms of information. Image and
video encryption have applications in various fields, including wireless
communications, multimedia systems, medical imaging, telemedi-
cine, and military communications [1–5].
In this book, image encryption algorithms are studied for the
­purpose of wireless communication of images in a secure form. The
main objective is to come to a conclusion whether a certain image
encryption algorithm is suitable for wireless communication applica-
tions. In this regard, we explore in this book some of the number-
theory-based encryption algorithms such as the Data Encryption
Standard (DES), the Advanced Encryption Standard (AES), and the
RC6 a­ lgorithms [6–12]. These algorithms are known in the literature
as strong encryption algorithms. Our main concern is not only the
strength of the encryption algorithm, but also its ability to work with
the limitations of wireless communication systems. Unfortunately,
some of the ciphers used currently were not designed for image
encryption. For example, the AES implemented in the Electronic
Code-Book (ECB) mode is not recommended for encrypting images
because of the repetitive data patterns existing in images. So, we
either modify these ciphers to work for image encryption or search for
other approaches suitable for this task. This book tries to investigate
both directions.
For this purpose, we have also explored another branch of cryp-
tography that depends on chaos theory. The usage of chaotic cryptog-
raphy for image encryption is promising. The idea of using chaos for

1
2 Im ag e En c ry p ti o n

encryption can be traced to Shannon’s classical article [13]. Chaotic

maps have been applied to cryptography in several different ways.
Chaotic sequences have several good properties, including ease of
generation, sensitive dependence on the initial conditions, and noise-
like behavior. Applying the chaos to cryptography was a great contri-
bution to improve the security of data due to the adequate properties
of chaotic sequences [14,15].
To make the study in this book comprehensive, we switch in
Chapters 7 and 8 to communication concepts concentrating on
the orthogonal frequency division multiplexing (OFDM) system.
In these chapters, we present a simplified model for the OFDM
communication system with its different implementations, with
­concentration on the limitations of this system, and how these limita-
tions affect the quality of received images after decryption. An exten-
sive c­ omparison study is presented in Chapter 9, considering different
possible s­cenarios and different encryption algorithms to determine
which encryption algorithm is suitable for which scenario. Finally, we
include all MATLAB® codes utilized in the experiments in this book
to open the door for researchers who wish to complete this study using
different ideas.
 2
Fundamentals of
I mag e E n cryp ti on

2.1 Introduction

We are living in the information age; we need to keep information

about every aspect of our lives. In other words, information is an asset
that has a value like any other asset. As an asset, information needs
to be secured from attacks. Since 1990, communication networks cre-
ated a revolution in the use of information. Authorized people can
send and receive information from a distance using ­communication
networks. To be secure, information needs to be hidden from unau-
thorized access (confidentiality), protected from unauthorized change
(integrity), and available to an authorized entity when it is needed
(availability). Although these three requirements have not changed,
they now have some new dimensions. Not only should information
be confidential when it is stored in a computer, but also there should
be a way to maintain its confidentiality when it is transmitted over a
communication network [16].
Information transmitted over computer networks currently is not
only text but also audio, images, and other multimedia types. The
field of multimedia security has matured in the past decade to provide
a class of tool sets and design insights for the protection and enhance-
ment of digital media under a number of diverse attack scenarios.
Research in multimedia security was first motivated partly by the
increasing use of digital means to communicate, store, and r­ epresent
entertainment information such as music and video. The digital form
allowed the perfect duplication of information and almost seam-
less manipulation and tampering of the data. This created new types
of security attacks not addressed in the past by the entertainment
industry. The paradigm shift from analog to digital m ­ ultimedia for

3
4 Im ag e En c ry p ti o n

entertainment has had an enormous impact on artists, publishers,

copyright holders, and consumers [17,18].
This chapter gives the principles underlying the design of crypto-
graphic algorithms. A review of the different issues that arise when
selecting, designing, and evaluating a cryptographic algorithm is pre-
sented. Diffusion-based cipher algorithms along with their modes of
operation are covered. Chaotic encryption with a Baker map is also
discussed.

2.2  Basic Concepts of Cryptography

Image security is based on cryptography. In fact, some basic concepts

from cryptography are used as building blocks (primitives) for appli-
cations in image security. For a better understanding of issues con-
cerning image security, an overview of cryptography is presented first.

2.2.1  Goals of Cryptography

Cryptography is a study of techniques (called cryptosystems) that are

used to accomplish the following four goals [1,16–18]:
• Confidentiality
• Data integrity
• Authentication
• Nonrepudiation
The study of the techniques used to break existing cryptosystems is
called cryptanalysis. Since cryptography and cryptanalysis are greatly
dependent on each other, people refer to cryptology as a joint study
of cryptography and cryptanalysis. Let us try to understand all four
goals of cryptography.
Confidentiality refers to the protection of information from unau-
thorized access. An undesired communicating party, called an adver-
sary, must not be able to access the communication material. This
goal of cryptography is a basic one that has always been addressed and
enforced throughout the history of cryptographic practice.
Data integrity ensures that information has not been manipulated in
an unauthorized way. If the information is altered, all ­communicating
parties can detect this alteration.
 F un da m en ta l s o f Im ag e En c ry p ti o n 5

Authentication methods are classified into two categories: entity

authentication and message authentication. Entity authentication is
the process by which one party is assured of the identity of a sec-
ond party involved in a protocol, and that the second has actually
­participated immediately prior to the time the evidence is acquired.
Message authentication is a term used analogously with data origin
authentication. It provides data origin authentication with respect to
the original message source and data integrity but with no uniqueness
and t­ imeliness guarantees.
Nonrepudiation means that the receiver can prove to everyone that
the sender did indeed send the message. That is, the sender cannot
claim that he or she did not encrypt or sign certain digital information.
Fortunately, modern cryptography has developed techniques to
handle all four goals of cryptography.

2.2.2  Principles of Encryption

The basic idea of encryption is to modify the message in such a

way that only a legal recipient can reconstruct its content [16–18].
A ­discrete-value cryptosystem can be characterized by
a set of possible plaintexts P
a set of possible ciphertexts C
a set of possible cipher keys K
a set of possible encryption and decryption transformations
E and D
An encryption system is also called a cipher, or a cryptosystem. The
message for encryption is called plaintext, and the encrypted message
is called ciphertext. Denote the plaintext and the ciphertext by P and
C, respectively [16–19]. The encryption procedure of a cipher can be
described as

C = E K e ( P ) (2.1)

where Ke is the encryption key, and E is the encryption function.

Similarly, the decryption procedure is defined as

P = DK d (C ) (2.2)

www.allitebooks.com
6 Im ag e En c ry p ti o n

Ke Kd
Recovered
Plaintext Ciphertext plaintext
Encryption Decryption
Public channel

Figure 2.1  Encryption/decryption of a cipher.

where Kd is the decryption key, and D is the decryption function.

The security of a cipher should only rely on the decryption key Kd
as an adversary can recover the plaintext from the observed cipher-
text once the adversary gets Kd. Figure 2.1 shows a block diagram for
­encryption/decryption of a cipher.

2.3  Classification of Encryption Algorithms

Encryption algorithms can be classified in different ways: according

to structures of the algorithms, according to keys, or according to the
percentage of the data encrypted [19,20].

2.3.1  Classification According to Encryption Structure

Encryption algorithms can be classified according to encryption

structure into block ciphers and stream ciphers.
A block cipher is a type of symmetric-key encryption algorithm
that transforms a fixed-length block of plaintext data into a block
of ciphertext data of the same length. The fixed length is called the
block size. For several block ciphers, the block size is 64 or 128 bits.
The larger the block size, the more secure is the cipher, but the more
complex are the encryption and decryption algorithms and devices.
Modern block ciphers have the following features [21]:
1. Variable key size
2. Mixed arithmetic operations, which can provide nonlinearity
3. Data-dependent rotations and key-dependent rotations
4. Lengthy key schedule algorithms
5. Variable plaintext/ciphertext block sizes and variable number
of rounds
Block ciphers can be characterized by
1. Block size: Larger block sizes mean greater security.
2. Key size: Larger key sizes mean greater security.
 F un da m en ta l s o f Im ag e En c ry p ti o n 7

3. Number of rounds: Multiple rounds increase security.

4. Encryption modes: They define how messages larger than the
block size are encrypted.
Unlike block ciphers that operate on large blocks of data, stream ciphers
typically operate on smaller units of plaintext, usually bits. So, stream
ciphers can be designed to be exceptionally fast, much faster than  a
typical block cipher. Generally, a stream cipher generates a sequence
of bits as a key (called key stream) using a pseudorandom number gen-
erator (PRNG) that expands a short secret key (e.g., 128 bits) into a
long string (key stream) (e.g., 106 bits), and the encryption is accom-
plished by combining the key stream with the plaintext. Usually, the
bitwise XOR operation is chosen to perform ciphering, basically for its
­simplicity [6–9]. Stream ciphers have the following properties [22, 23]:
1. They do not have perfect security.
2. Security depends on the properties of the PRNG.
3. The PRNG must be unpredictable; given a consecutive
sequence of output bits, the next bit must be hard to predict.
4. Typical stream ciphers are very fast.

2.3.2  Classification According to Keys

According to keys, there are two kinds of ciphers following the

­relationship of Ke and Kd. When Ke = Kd, the cipher is called a
­private-key cipher or a symmetric cipher. For private-key ciphers, the
encryption/decryption key must be transmitted from the sender to
the receiver via a separate secret channel. When Ke ≠ Kd, the cipher
is called a public-key cipher or an asymmetric cipher. For public-key
ciphers, the encryption key Ke is published, and the decryption key Kd
is kept private, for which no additional secret channel is needed for
key transfer. In conventional encryption as shown in Figure 2.2, the
sender encrypts the data (plaintext) using the encryption key, and the
receiver decrypts the encrypted data (ciphertext) into the original data
(plaintext) using the decryption key. In symmetric encryption, both
encryption and decryption keys are identical. Figure  2.3 shows the
public-key encryption (asymmetric encryption), in which the encryp-
tion and decryption keys are different. Public-key cryptography solves
the problem of conventional cryptosystems by distributing the key
8 Im ag e En c ry p ti o n

Sender Transmission Receiver

Xe Kd
Message Message
Source Destination
Cryptanalyst
X (Attacks)
X
Encryption Decryption
Algorithm Y Y Algorithm

K
K
Secret Secure
Key Channel

Figure 2.2  Model of symmetric encryption.

X
Cryptanalyst .G
The Source A
The Destination B
Message X Y X
Encrypt Decrypt Destination
source

.G
.H

Key pair
source

Figure 2.3  Asymmetric-key encryption.

[24–27]. Table 2.1 shows a comparison between symmetric encryp-

tion and asymmetric encryption.
In general, there are two types of cryptosystems: (1) ­symmetric-
(private-) key cryptosystems and (2) asymmetric- (public-) key cryp-
tosystems. Most people have chosen to call the first group simply
symmetric-key cryptosystems, and the popular name for the second
group is just public-key cryptosystems.

2.3.3  Classification According to Percentage of Encrypted Data

With respect to the amount of encrypted data, the encryption can

be divided into full encryption and partial encryption (also called
selective encryption) according to the percentage of the data
encrypted.
 F un da m en ta l s o f Im ag e En c ry p ti o n 9

Table 2.1  Comparison between Symmetric Encryption and Asymmetric Encryption

CONVENTIONAL ENCRYPTION (SYMMETRIC PUBLIC-KEY ENCRYPTION (ASYMMETRIC
ENCRYPTION) ENCRYPTION)
Requirements to work: Requirements to work:
1. The same algorithm with the same key can 1. One algorithm is used for encryption and
be used for encryption and decryption. decryption with a pair of keys, one for
2. The sender and receiver must share the encryption and one for decryption.
algorithm and the key. 2. The sender and receiver must each have
one of the matched pair of keys.
Requirements for security: Requirements for security:
1. The key must be kept secret. 1. The decryption key must be kept secret.
2. It must be impossible or at least 2. It must be impossible or at least
impractical to decipher a message if no impractical to decipher a message if no
other information is available. other information is available.
3. Knowledge of the algorithm plus samples 3. Knowledge of the algorithm, the encryption
of the ciphertext must be insufficient to key, and samples of the ciphertext must be
determine the key. insufficient to determine the decryption key.

2.4 Cryptanalysis

Cryptanalysis is the art of deciphering an encrypted message as a

whole or in part when the decryption key is not known. Depending
on the amount of known information and the amount of control over
the  system by the adversary (cryptanalyst), there are several basic
types of cryptanalytic attacks [28–30]. Some of the most important
ones, for a system implementer, are described next, and they are sum-
marized in Table 2.2.
Ciphertext-only attack. The adversary only has access to one or
more encrypted messages. The most important goal of a pro-
posed cryptosystem is to withstand this type of attack.
Brute-force attack. This is a type of ciphertext-only attack. It is
based on an exhaustive key search, and for well-designed
cryptosystems, it should be computationally infeasible.
Known-plaintext attack. In this type of attack, an adversary has
some knowledge about the plaintext corresponding to the
given ciphertext. This may help the adversary determine the
key or a part of the key.
Chosen-plaintext attack. Essentially, an adversary can feed the
chosen plaintext into the black box that contains the encryp-
tion algorithm and the encryption key. The black box gives
the corresponding ciphertext, and the adversary may use the
10 Im ag e En c ry p ti o n

Table 2.2  Types of Attacks on Encrypted Images

TYPE OF ATTACKS PREREQUISITES FOR THE CRYPTANALYST
Ciphertext only 1. Encryption algorithm
2. Ciphertext to be decoded
Known plaintext 1. Encryption algorithm
2. Ciphertext to be decoded
3. Plaintext message together with its corresponding ciphertext
generated with the secret key
Chosen ciphertext 1. Encryption algorithm
2. Ciphertext to be decoded
3. Reported ciphertext chosen by cryptanalyst together with its
corresponding plaintext generated with the decryption
algorithm and the decryption key
Chosen plaintext 1. Encryption algorithm
2. Ciphertext to be decoded
3. Reported plaintext chosen by cryptanalyst together with its
corresponding ciphertext generated with the encryption
algorithm and the encryption key

accumulated knowledge about the plaintext–ciphertext pairs

to obtain the secret key or at least a part of it.
Chosen-ciphertext attack. Here, an adversary can feed the cho-
sen ciphertext into the black box that contains the decryption
algorithm and the decryption key. The black box produces the
corresponding plaintext, and the adversary tries to obtain the
secret key or a part of the key by analyzing the accumulated
ciphertext–plaintext pairs.

2.5  Features of Image Encryption Schemes

Unlike text messages, image data have their special features, such as
high redundancy and high correlation among pixels. Also, they are usu-
ally huge in size. Together, these make traditional encryption methods
difficult to apply and slow to process. Sometimes, image applications
have their own requirements, like real-time processing, fidelity reser-
vation, image format consistency, data compression for transmission,
and so on. Simultaneous fulfillment of these requirements along with
high-security and high-quality demands has presented great chal-
lenges to real-time imaging practice. For studying image encryption,
 F un da m en ta l s o f Im ag e En c ry p ti o n 11

we must first analyze the differences between implementations for

image data and text data encryption. Basically, there are some differ-
ences between image and text data encryption [31–34]:
1. When the ciphertext is produced, it must be decrypted to
the original plaintext in a full lossless manner. However, the
cipherimage can be decrypted to the original plainimage in
some lossy manner.
2. Text data are sequences of words. They can be encrypted
directly by using block or stream ciphers. However, digi-
tal images are usually represented as two-dimensional (2D)
arrays. For protecting the stored 2D arrays of data with text-
processing algorithms, they must be converted to 1D arrays
before using various traditional encryption techniques.
3. Because the storage space of a picture is very large, it is some-
times inefficient to encrypt or decrypt images directly. One
of the best methods is to encrypt/decrypt information that is
used by image compression only for reducing both its storage
space and transmission time.

2.6  Conventional Symmetric Block Ciphers

This section gives a brief overview of the construction of some popular

conventional encryption algorithms. Each of the following encryp-
tion algorithms is a symmetric block cipher algorithm. Symmetric
means that the key used for encryption and decryption is the same;
block means that the data (information) to be encrypted is divided into
blocks of equal length [35–45].

2.6.1  Data Encryption Standard

The Data Encryption Standard (DES) is the most well-known sym-

metric-key block cipher. It was selected by the National Bureau of
Standards as an official Federal Information Processing Standard
(FIPS) for the United States in 1976, and it subsequently enjoyed
widespread use internationally [46].
The DES is a block cipher, which encrypts data in 64-bit blocks.
A 64-bit block of plaintext goes at one end of the algorithm, and
a 64-bit block of ciphertext comes out at the other end. The same
12 Im ag e En c ry p ti o n

32 1 2 3 4 5 6 7 8 9 10 11 12 32

48 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 48

Figure 2.4  Expansion process.

algorithm and key of size 56 bits are used for both encryption and
decryption except for minor differences in the key schedule. The key
is usually expressed as a 64-bit number, but every eighth bit, one bit
is used for parity checking and is ignored. These parity bits are the
least-significant bits of the key bytes. The key can be any 56-bit num-
ber and can be changed at any time, although some selections can be
considered weak keys.
The DES is based on four basic operations: expansion, permutation,
XOR, and substitution. The data to be encrypted are first divided into
64-bit blocks and fed into an Initial Permutation (IP) stage, in which
each block is divided into two subblocks, each with a 32-bit length.
The right subblock is fed into a Feistel function (f-function), which is
depicted in Figure 2.4. It operates on half a block (32 bits) at a time
and consists of four stages as shown in Figure 2.5.
1. Expansion. The 32-bit half block is expanded to 48 bits using
the expansion permutation, denoted as E in the diagram, by
duplicating half of the bits. The output consists of eight 6-bit
(8 × 6 = 48 bits) pieces, each containing a copy of 4 corre-
sponding input bits plus a copy of the immediately adjacent
bit from each of the input pieces to either side.
2. Key mixing. The result is combined with a subkey using
an XOR operation. Sixteen 48-bit subkeys, one for each
round, are derived from the main key using a key-schedule
mechanism.
3. Substitution. After mixing with the subkey, the block is
divided into eight 6-bit pieces before processing by the
Substitution boxes (S-boxes). Each of the eight S-boxes
replaces its six input bits with four output bits according to
a nonlinear transformation, provided in the form of a lookup
table. The S-boxes provide the core of security of the DES.
 F un da m en ta l s o f Im ag e En c ry p ti o n 13

Plaintext 64 bits
IP

For 16 rounds

FP
Ciphertext 64 bits

Half block (32 bits) Sub-key (48 bits) f-function

S S S S S S S S

Figure 2.5  The DES algorithm and f-function.

Without them, the cipher would be linear and trivially

breakable.
4. Permutation. Finally, the 32 outputs from the S-boxes are
rearranged according to a fixed permutation, the P-box,
which is designed so that, after expansion, each group of
S-box output bits is spread across six different S-boxes in the
next round.
At its simplest level, the DES algorithm is nothing more than a
combination of the two basic techniques of encryption: confusion
14 Im ag e En c ry p ti o n

and diffusion. The fundamental building block of the DES is a

­single combination of these techniques: a substitution followed
by a permutation on the data based on the key. After an initial
­permutation, each block is broken into a right half and a left half,
each with 32 bits. Then, there are 16 rounds of identical opera-
tions, called f functions, in which the data are combined with the
key. After these rounds, the right and left halves are joined, and a
final permutation, the inverse of the initial permutation, finishes
the algorithm.

2.6.2  Double DES

A naive way of improving the security of a block cipher algorithm is

to encrypt each block twice with two different keys. First, encrypt a
block with the first key and then encrypt the resulting ciphertext with
the second key. Decryption is the reverse process. In the Double DES
encryption algorithm, each 64-bit block of data is encrypted twice
with the DES algorithm, first with a key K1 and then with another
key K 2. The scheme involves a key of 112 bits.
The resultant doubly encrypted ciphertext block should be
much harder to decrypt using an exhaustive search. Instead of 256
attempts, it requires 2128 attempts to find the key and 2112 attempts
to break the encryption. In 1981, Merkle and Hellman declared
their “­meet-in-the-middle” attack, which proved the weakness of
the Double DES algorithm [47]. The “meet-in-the-middle” attack
is a known plaintext attack that requires that an attacker have
both a known piece of plaintext and the corresponding ciphertext.
The attack requires storing 256 intermediate results when trying to
crack a message that has been encrypted with the double DES.
Merkle and Hellman developed a time-memory trade-off that
could break this d­ ouble-DES encryption scheme in 256+1 trials, not
in 2112 trials.

2.6.3  Triple DES

The dangers of the Merkle-Hellman “meet-in-the-middle” attack can

be circumvented by performing three block encryption operations.
 F un da m en ta l s o f Im ag e En c ry p ti o n 15

This method is called Triple DES. Triple DES is performed by exe-

cuting the DES three times, producing an effective key size of 168
bits. In the Triple DES, each 64-bit block of data is encrypted three
times with the DES algorithm. In practice, the most common way to
perform the Triple DES is
1. Encrypt with key 1.
2. Decrypt with key 2.
3. Encrypt with key 3.
To decrypt, reverse the steps:
1. Decrypt with key 3.
2. Encrypt with key 2.
3. Decrypt with key 1.
For several applications, we can use the same key for both key 1 and
key 3 without creating a significant vulnerability. The choice between
Single, Double, and Triple DES is a trade-off between performance
and security requirements [45].

2.6.4  International Data Encryption Algorithm

The International Data Encryption Algorithm (IDEA) cipher was

first presented by Lai and Massey in 1990 under the name Proposed
Encryption Standard (PES). After Biham and Shamir presented
differential cryptanalysis, the authors strengthened their cipher
against the attack and named it the Improved Proposed Encryption
Standard (IPES). The IPES name was changed to the International
Data Encryption Algorithm (IDEA) in 1992. This algorithm was
intended as a replacement for the DES. IDEA is a block cipher that
operates on 64-bit plaintext blocks. The key is 128 bits long. There
are eight identical rounds, and the same algorithm is used for both
encryption and decryption. It uses both confusion and diffusion.
The design philosophy behind the algorithm is based on mixing
operations from different algebraic groups. Three algebraic groups
(XOR, addition modulo, and multiplication modulo) are mixed in
this algorithm [16,18], and they are all easily implemented in both
hardware and software. All of these operations operate on 16-bit
subblocks.

www.allitebooks.com
16 Im ag e En c ry p ti o n

2.6.5 Blowfish

Blowfish is a 64-bit block cipher with a variable-length key. It was

designed in 1993 by Schaneier [16]. This algorithm consists of two
parts: key expansion and data encryption. Key expansion converts a
key of up to 448 bits into several subkey arrays totaling 4168 bytes.
Data encryption consists of a simple function iterated 16 times.
Each round consists of a key-dependent permutation and a key-
and data-dependent substitution. All operations are additions and
XORs on 32-bit words. The only additional operations are four
indexed array data lookups per round. The keys must be precom-
puted before any data encryption or decryption, with decryption
exactly the same as encryption except that the subkeys are used in
reverse order.

2.6.6  RC5 Algorithm

The iterated block RC5 was introduced by Rivest, Shamir, and

Adleman in 1994 [48]. The main feature of the RC5 is the heavy
use of data-dependent rotations. RC5 has a variable word size w, a
variable number of rounds r, and a variable secret key with b bytes.
It is represented as RC5 w/r/b. The nominal value of w is 32 bits, and
RC5 encrypts blocks of two words. The RC5 is composed of encryp-
tion, decryption, and key expansion. The expanded key ­ contains
t = 2 × (r + 1) words. The primitive operations of the RC5 are illustrated
in Table 2.3. Generally, RC5 is a fast symmetric block cipher that is
suitable for hardware and software implementations with low mem-
ory requirements. It provides high security when good p ­ arameters are
chosen.

Table 2.3  Primitive Operations of RC5

a+b Integer addition modulo 2w
a–b Integer subtraction modulo 2w
a⊕b Bitwise XOR of w-bit words
a*b Integer multiplication modulo 2w
a <<< b Rotate the w-bit word a to the left by the amount given by the
least-significant lg w bits of b
a >>> b Rotate the w-bit word a to the right by the amount given by the
least-significant lg w bits of b
 F un da m en ta l s o f Im ag e En c ry p ti o n 17

Ai–1 Bi–1

+ +

<<< <<<

+ S2i
+ S2i+1

Ai Bi

Figure 2.6  RC5w/r/b symmetric block cipher diagram.

2.6.6.1  RC5 Encryption Algorithm  We assume that the input block is

given in two w-bit registers A and B, and we also assume that key
expansion has already been performed so that S[0], S[1], …, S[t - 1]
have been computed. The steps of the encryption algorithm can be
summarized as follows:

A = A+S [0];
B = B+S [1];
For i = 1 to r do
A = ((A ⊕ B) <<< B) +S [2i];
B = ((B ⊕ A) <<< A) +S [2i+1];
End

In each round of RC5, both registers A and B are updated as shown

in Figure 2.6.

2.6.6.2  RC5 Decryption Algorithm  The decryption step can be sum-

marized as follows:

For i = r downto 1 do
B = ((B - S [2i+1]) >>> A) ⊕ A;
A = ((A - S [2]) >>> B) ⊕ B;
End
B = B-S [1];
A = A-S [0];

2.6.6.3  RC5 Key Expansion  Key expansion expands the user’s secret
key K to fill the expanded key array S, which makes S similar to an
18 Im ag e En c ry p ti o n

array of t = 2(r + 1) random binary words. Two magic constants, Pw

and Qw, are used in this process. These constants are defined as

( )
Pw = Odd ( e − 2 ) 2w (2.3)

( )
Qw = Odd ( φ − 1) 2w (2.4)

where
e = 2.718281828459....(base of natural logarithms)
ϕ = 1.618033988749....(golden ratio)
and Odd(x) is the odd integer nearest to x. For w = 16 and 32, these
constants are given in hexadecimal:

P16 = b7e1; Q16 = 9e37

P32 = b7e15163; Q32 = 9e3779b9,

The expansion begins by copying the secret key K[0....b-1] into an

array L[0....c-1] that has c = [ b / u ] words, where u = w/8 is the number
of bytes per word. u consecutive key bytes of K are used to fill each
successive word in L in a low-order to high-order byte manner. All
unfilled byte positions of L are zeroed.
To initialize the array S, we use the following steps:

S [0] = PW;
For i = 1 to t-1 do
S [i] = S [i-1] + QW;
End

The last step is to mix the user secret key in three passes over the
arrays S and L as follows:

i = j = 0;
A = B = 0;
Do 3*max (t, c) times:
A = S [i] = (S [i] +A+B) <<<3;
B = L[j] = (L[j] + A+ B) <<< (A+B);
i = (i+1) mod (t);
j = (j+1) mod (c);
 F un da m en ta l s o f Im ag e En c ry p ti o n 19

2.6.7  RC6 Algorithm

The RC6 block cipher is a modified version of RC5 that uses four
working registers instead of two and integer multiplication as an addi-
tional primitive operation. The integer multiplication process greatly
enhances the diffusion achieved per round, which leads to greater
security, fewer rounds, and increased throughput. The key schedule
of RC6-w/r/b is similar to the key schedule of RC5-w/r/b. The only
difference is that for RC6-w/r/b, more words are derived from the
user-supplied key for use during encryption and decryption. The user
supplies a key of b bytes, where 0 ≤ b ≤ 255. From this key, 2r + 4
words (w bits each) are derived and stored in the array S[0, … , 2r + 3].
This array is used in both encryption and decryption [12]. Generally,
RC6 consists of two Feistel networks whose data are mixed via data-
dependent rotations. The operations in a single round of RC6 con-
tain two applications of the squaring function f(x) = x(2x + 1) mod
232, two fixed 32-bit rotations, two data-dependent 32-bit rotations,
two XORs, and two additions modulo 232. The steps of RC6 encryp-
tion and decryption are summarized next, and the block diagrams of
RC6 encryption and decryption are shown in Figures  2.7 and 2.8,
respectively.

2.6.7.1 RC6 Encryption Algorithm

Input: Four w-bit plaintext values stored in registers

A, B, C, and D
Number r of rounds
w-bit round keys S[0, …, 2r+3]
Output: Four w-bit ciphertext values stored in
registers A, B, C, and D.
Procedure: B = B + S [0];
D = D + S [1];
For i = 1 to r do
{t = (B × (2B + 1)) <<< lg w;
u = (D × (2D + 1)) <<< lg w;
A = ((A ⊕ t) <<< u) + S [2i];
C = ((C ⊕ u) <<< t) + S [2i + 1];
(A, B, C, D) = (B, C, D, A);}
End
A = A + S [2r + 2];
C = C + S [2r + 3];
20 Im ag e En c ry p ti o n

A B C D

+ S[0] + S[1]

+ <<< f + <<< f

lg w lg w
<<<

<<<

Repeat for r rounds

+ S[2i] + S[2i+1]

+ S[2i+2] + S[2r+3]

A B C D

Figure 2.7  Encryption with RC6-w/r/b algorithm.

2.6.7.2  RC6 Decryption Algorithm

Input: Four w-bit ciphertext values stored in registers

A, B, C, and D
Number of rounds r
w-bit round keys S[0, …, 2r+3]
Output: Four w-bit plaintext values stored in
registers A, B, C, and D.
Procedure: C = C − S [2r + 3];
A = A − S [2r + 2];
for i = r downto 1 do
{(A, B, C, D) = (D, A, B, C);
u = (D × (2D + 1)) <<< log(w);
t = (B × (2B + 1)) <<< log(w);
C = ((C − S[2i + 1]) >>> t) ⊕ u;
 F un da m en ta l s o f Im ag e En c ry p ti o n 21

A B C D

– S[2r+2] – S[2r+3]

– S[2i] – S[2i+1]

Repeat for r rounds

>>>

>>>

+ <<< f + <<< f

log(w) log(w)

– S[0] – S[1]

A B C D

Figure 2.8  Decryption with RC6-w/r/b algorithm.

A = ((A − S[2i]) >>> u) ⊕ t;}

End
D = D − S [1];
B = B − S [0];

2.6.8  The Advanced Encryption Standard

The Advanced Encryption Standard (AES) is based on the Rijndael

algorithm, which is an iterated block cipher algorithm with a variable
block size and a variable key size. The block size and the key size can
be independently 128, 192, or 256 bits. The intermediate resulting
ciphertext is called a state, and it is in the form of a rectangular array
of four rows and a number of columns equal to the block size divided
22 Im ag e En c ry p ti o n

by 32. The cipher key is similarly a rectangular array with four rows
and a number of columns equal to the key size divided by 32. The
number of rounds performed on the intermediate state is related to the
key size. For key sizes of 128, 192, and 256 bits, the number of rounds
is 10, 12, and 14, respectively. Each round consists of a fixed sequence
of transformations, except the first and the last round [16–18].
The AES consists of rounds. Any round, except the final one,
consists of SubBytes, ShiftRows, MixColumns, and AddRoundKey
operations. In the final round, no MixColumns operation is per-
formed. In the SubBytes step, a linear substitution for each byte is
performed according to Figure 2.9. Each byte in the array is updated
using an 8-bit S-box, which provides the nonlinearity in the cipher
system [16–18].
The S-box is derived from the multiplicative inverse over the finite
Galois field GF(28), known to have good nonlinearity properties. To
avoid attacks based on simple algebraic properties, the S-box is cho-
sen to avoid any fixed points and any opposite fixed points [16–18].
The ShiftRows step operates on the rows of the state. It cycli-
cally shifts the bytes in each row. For the AES, the first row is left
unchanged. Each byte of the second row is shifted a single byte to the
left. Similarly, the third and fourth rows are shifted by offsets of 2 and
3 bytes, respectively. For the block of size 128 bits and 192 bits, the
shifting pattern is the same [16–18].
In this way, each column of the output state of the ShiftRows step
is composed of bytes from each column of the input state. In the case
of the 256-bit blocks, the first row is unchanged, and the shifting for

b1 b2 b3 b4 d1 d2 d3 d4

b5 b6 b7 b8 SubBytes d5 d6 d7 d8

b9 b10 b 11 b12 d9 d10 d 11 d12

b13 b14 b15 b16 d13 d14 d15 d16

Figure 2.9  SubBytes step.

 F un da m en ta l s o f Im ag e En c ry p ti o n 23

b1 b2 b3 b4 b1 b2 b3 b4

b5 b6 b7 b8 b6 b7 b8 b5
ShiftRows

b9 b10 b11 b12 b10 b11 b12 b9

b13 b14 b15 b16 b14 b15 b16 b13

Figure 2.10  ShiftRows step.

b1 b2 b3 b4 d1 d2 d3 d4

b5 b6 b7 b8 d5 d6 d7 d8
MixColumns

b9 b10 b11 b12 d9 d10 d11 d12

b13 b14 b15 b16 d13 d14 d15 d16

C(x)

Figure 2.11  MixColumns step.

second, third, and fourth rows is 1 byte, 3 bytes, and 4 bytes, respec-
tively, as shown in Figure 2.10.
In the MixColumns step, the 4 bytes of each column of
the  state are combined using an invertible linear transforma-
tion.  The  MixColumns function takes 4 bytes as input and out-
puts 4 bytes, where each input byte affects all 4 output bytes. With
ShiftRows, MixColumns provides diffusion in the cipher system.
Each column is treated as a p ­ olynomial over GF(28) and is then
multiplied with a fixed polynomial C(x) = 3x 3 + x 2 + x + 2. The
MixColumns step can also be viewed as multiplication by a particu-
lar matrix as shown in Figure 2.11 [16–18].
In the AddRoundKey step, the subkey is combined with the state.
For each round, a subkey is derived from the main key using the
algorithm key schedule. Each subkey has the same size as the state.
24 Im ag e En c ry p ti o n

b1 b2 b3 b4 d1 d2 d3 d4

b5 b6 b7 b8 d5 d6 d7 d8
AddRoundKey

b9 b10 b11 b12 d9 d10 d11 d12

b13 b14 b15 b16 d13 d14 d15 d16

C(x)

k1 k2 k3 k4

k5 k6 k7 k8

k9 k10 k11 k12

k13 k14 k15 k16

Figure 2.12  AddRoundKey step.

The subkey is added by combining each byte of the state with the
corresponding byte of the subkey using a bitwise XOR [16–18]. The
AddRoundKey step is shown in Figure 2.12. We apply the AES with
a fixed block size of 128 bits and a key size of 128 bits.

2.7  Modes of Operation

Block ciphers can be run in different modes of operation, allowing users

to choose appropriate modes to meet the requirements of their appli-
cations. Using a certain mode in the encryption process restricts the
decryption process to use the same mode. In this section, we discuss dif-
ferent possible ways in which block codes can be utilized to implement a
cryptosystem. The possible block cipher modes of operation that we treat
are identified by the acronyms electronic codebook (ECB), cipher-block
chaining (CBC), cipher feedback (CFB), and output feedback (OFB). In
each case, we assume that we have a block cipher of block length n with
enciphering map EK and deciphering map DK for each key K.
 F un da m en ta l s o f Im ag e En c ry p ti o n 25

2.7.1  The ECB Mode

ECB is the simplest mode of operation for encryption algorithms;

the data sequence is divided into blocks of equal size, and each block
is encrypted, separately, with the same encryption key. As illustrated
in Figure 2.13, the plaintext is divided into blocks (P1, P2, P3, ....) of
size n bits that are encrypted to ciphertext blocks (C1, C2, C3, ....). The
encryption algorithm is

Cj = EK (Pj) (2.5)

and the decryption algorithm is

Pj = DK (Cj) (2.6)

where j = 1,  2,  3, ... ; EK is the encryption map with the key K; and DK
is the decryption map with the same key K.
The ECB mode has several advantages. There is no need to encrypt
a file progressively; the middle blocks can be encrypted first, then
the blocks at the end, and finally the blocks at the beginning. This is
important for encrypted files that are accessed randomly, like a data-
base. If a database is encrypted in the ECB mode, then any record can
be added, deleted, encrypted, or decrypted independently, assuming
that a record consists of independent encryption blocks.
The disadvantage of this mode is that identical plaintext blocks
are encrypted to identical ciphertext blocks; it does not hide data
patterns. The advantage is that error propagation is limited to a
single block. The disadvantage of the ECB mode appears in image
encryption if there is an image with large areas of the same color or
repeated patterns so that there are many blocks of the same plaintext.

Plaintext Plaintext Plaintext

P1 P2 P3

Encryption Encryption Encryption

Key Key Key
Algorithm Algorithm Algorithm

Ciphertext Ciphertext Ciphertext

C1 C2 C3

Figure 2.13  Using a block cipher in the ECB mode.

www.allitebooks.com
26 Im ag e En c ry p ti o n

This may reveal much information about the original image from the
encrypted image. This disadvantage is treated in CBC, CFB, and
OFB modes.

2.7.2  The CBC Mode

The CBC mode uses an initialization vector (IV) of size equal to the
size of each block of pixels. In this mode, each block of plaintext is
XORed with the previous ciphertext block before being encrypted.
By this way, each ciphertext block is dependent on all plaintext
blocks up to that point. In decryption, the same XOR operation is
repeated so that its effect is cancelled. This mechanism is shown in
Figure 2.14.
The main disadvantage of the CBC mode is that an error in (or
attack on) one ciphertext block impacts two plaintext blocks on
decryption. On the other hand, if there is an image that has blocks
of the same input data, these blocks are encrypted to totally different
ciphertext data. So, the CBC mode is a better approach in encrypting
images in the spatial domain, especially when these images contain
large areas of the same activity. In the CBC mode, the encryption
algorithm is

Cj = EK (Cj−1 ⊕ Pj) (2.7)

and the decryption algorithm is

Pj = DK (Cj) ⊕ Cj−1, j = 1, 2, 3, ... (2.8)

C 0 = IV (2.9)

P1 P2 P3

C0 = IV + + +

Key EK Key EK Key EK

C1 C2 C3

Figure 2.14  Using a block cipher in the CBC mode.

 F un da m en ta l s o f Im ag e En c ry p ti o n 27

2.7.3  The CFB Mode

In contrast to the CBC mode, the CFB mode begins by encrypt-

ing the IV, and then an XOR operation is performed between the
bits of the encrypted IV and the corresponding bits of the first block
of the image. The result is the encrypted version of the first block.
For the encryption of each of the next plaintext blocks, the previous
ciphertext block is encrypted, and the output is XORed with the cur-
rent plaintext block to create the current ciphertext block. The XOR
operation conceals plaintext patterns.
Common to the CBC mode, changing the IV to the same plain-
text block results in different outputs. Although the IV need not be
secret, some applications would see this as desirable [16–18,49,50].
Figure 2.15 shows the CFB mode. The encryption algorithm is

Cj = Pj ⊕ Ij (2.10)

and the decryption algorithm is

Pj = Cj ⊕ Ij (2.11)

Ij = EK (Cj-1), j = 1, 2, 3, ... (2.12)

C 0 = IV (2.13)
2.7.4  The OFB Mode

The OFB mode is similar to the CFB mode. It begins by e­ ncrypting

the IV. The bits of the encrypted IV are XORed with the correspond-
ing bits of the first plaintext block to obtain the corresponding cipher-
text block. Also, the output of the encryption algorithm is used as
an input to the next encryption step instead of the IV. This process

C0 = IV EK I1 EK I2 EK I3

P1 P2 P3

+ + +

C1 C2 C3

Figure 2.15  Using a block cipher in the CFB mode.

28 Im ag e En c ry p ti o n

I0 = IV EK I1 EK I2 EK I1

P1 P2 P3

+ + +

C1 C2 C3

Figure 2.16  Using a block cipher in the OFB mode.

continues until the last block. Changing the IV for the same ­plaintext
block results in different ciphertext blocks. Figure  2.16 shows the
OFB mode [16–18,49,50]. The encryption algorithm is
Cj = Pj ⊕ Ij (2.14)
and the decryption algorithm is

Pj = Cj ⊕ Ij (2.15)
Ij = EK(Ij-1), j = 1,2,3,..... (2.16)
I0 = IV (2.17)

2.8  Chaos and Cryptography

Chaos theory has been established since the 1970s in many different
research areas, such as physics, mathematics, engineering, biology,
and others. The most well-known characteristics of chaos are the so-
called butterfly effect (sensitivity to initial conditions) and the pseudo-
randomness generated by deterministic equations. Many fundamental
properties of chaotic systems have their corresponding counterparts
in traditional cryptosystems. Chaotic systems have several significant
features favorable to secure communications, such as ergodicity, sen-
sitivity to initial conditions and control parameters, and random-like
behavior [38–40]. With all these advantages, scientists expected to
introduce new and powerful tools of chaotic cryptography. So, chaos
has become a new rich source of new ciphers [31–35].
Chaotic dynamic systems are dimensional nonlinear dynamic sys-
tems that are capable of complex and unpredictable behavior. Chaos
describes a system that is sensitive to initial conditions to generate an
apparently random behavior but at the same time it is completely deter-
ministic. These properties of chaos have much p ­ otential for ­applications
 F un da m en ta l s o f Im ag e En c ry p ti o n 29

in c­ ryptography as it is hard to make long-term predictions on c­ haotic

systems. First, being completely deterministic means that we can
always obtain the same set of values provided we have exactly the same
mapping function and initial conditions. Because chaotic functions
are sensitive to initial conditions, any slight difference in the initial
values used means that the ciphertext produced using chaos will be
completely different. This means that the system will be strong against
brute-force attacks as the number of possible keys is large.
The basic properties of chaotic systems are the deterministicity, the
sensitivity to initial conditions and parameters, the topological transi-
tivity, and the ergodicity [51–55].
Deterministicity means that chaotic systems have some determin-
ing mathematical equations ruling their behavior.
Sensitivity to initial conditions means that when a chaotic map is
iteratively applied to two initially close points, the iterations quickly
diverge and become uncorrelated in the long term. Sensitivity to
parameters causes the properties of the map to change quickly when
slightly perturbing the parameters on which the map depends. Hence,
a chaotic system can be used as a pseudorandom number generator.
Topological transitivity is the tendency of the system to quickly
scramble up small portions of the state space into an intricate network
of filaments. Local, correlated information becomes scattered over the
state space.
The ergodicity of a chaotic map means that if the state space is
partitioned into a finite number of regions, no matter how many, any
orbit of the map will pass through all these regions. The ergodicity of
a chaotic map is ensured by the topological transitivity property.
Many properties of chaotic systems, such as mixing and sensi-
tivity to initial conditions, have their corresponding ­counterparts
in traditional cryptosystems. Table  2.4 contains a partial list  of

Table 2.4  Similarities and Differences between Chaotic Systems and Cryptographic Algorithms
CHAOTIC SYSTEMS CRYPTOGRAPHIC ALGORITHMS
Set of real numbers Finite set of integers
Iterations Rounds
Parameters Key
Sensitivity to a change in initial conditions Diffusion
— Security
30 Im ag e En c ry p ti o n

these properties. The main difference between chaos theory and

­cryptography is that ­cryptosystems work on a finite field, while chaos
works on a continuum.
Several chaos-based image encryption schemes been developed in
recent years. In 1992, Bourbakis and Alexopoulos proposed an image
encryption scheme that utilizes the SCAN language to encrypt and
compress an image simultaneously [56]. Fridrich demonstrated the
construction of a symmetric block cipher algorithm based on a 2D
standard Baker map [52]. There are three basic steps in Fridrich’s
method [52]: Choosing a chaotic map and generalizing it by introduc-
ing some parameters, discretizing the chaotic map to a finite square
lattice of points that represent pixels, and extending the discredited
map to three dimensions and further composing it with a simple dif-
fusion mechanism.
Scharinger designed a chaotic Kolmogorov-flow-based image
encryption technique in which the whole image is taken as a single
block and permuted through a key-controlled chaotic system [57]. In
addition, a shift register pseudorandom generator is adopted to intro-
duce the confusion in the data. Yen and Guo proposed an encryption
method called bit recirculation image encryption (BRIE) based on a
chaotic logistic map [58]. The basic principle of this BRIE method is
bit recirculation of pixels, which is controlled by a chaotic pseudoran-
dom binary sequence. The secret key of the BRIE method consists of
two integers and an initial condition of the logistic map. Yen and Guo
[59] also proposed an encryption method called CKBA (chaotic key-
based algorithm), in which the key binary sequence is generated using
a chaotic system. The image pixels are rearranged according to the
generated binary sequence and then XORed and XNORed with the
selected key. In 2002, Li and Zheng [29] pointed out some defects in
the encryption schemes presented in the references [58] and discussed
some possible improvements for them. Chen et al. [60] proposed a
symmetric algorithm in which a 2D chaotic map is generalized to
three dimensions for designing a real-time secure image encryption
scheme. This approach employs the 3D cat map to shuffle the posi-
tions of the image pixels and uses another chaotic map to confuse the
relationship between the encrypted image and its original image.
In chaotic encryption with 1D maps, the encryption key is gener-
ated with a chaotic map based on selected initial conditions. Each
 F un da m en ta l s o f Im ag e En c ry p ti o n 31

map produces various random keys from various orbits of the map,
which guarantees the security. Based on the key, a binary sequence
is generated to control the encryption algorithm. The input image
of two dimensions is transformed into a 1D array and then divided
into various subblocks. Then, the position permutation and value per-
mutation are applied to each binary matrix representing a subblock.
Finally, the receiver uses the same subkeys to decrypt the encrypted
images. On the other hand, chaotic encryption with 2D maps like the
Baker map is different, as shown in the next section.

2.9  The 2D Chaotic Baker Map

The Baker map stretches the image horizontally and then folds it
vertically. Repeating this process, the positions of all pixels of the
plainimage are changed [51,52,61]. Let B(n1, … , nk ) denote the dis-
cretized map, where the vector [n1, … , nk] represents the secret key
Skey. Defining N as the number of data items in one row, the secret key
is chosen such that each integer ni divides N, and n1 + … + nk = N.
Let Ni = n1 + … + ni. The data item at the indices (q, z) is moved to
the indices:

N N
B(n1 ,...,nk ) ( q , z ) =  ( q − N i ) + z mod   ,
 ni  ni 

ni   N  
 z − z mod   + N  (2.18)
 ni  
i
N 

where Ni ≤ q < Ni + ni, and 0 ≤ z < N.

In steps, the chaotic permutation is performed as follows:
1. An N × N square image is divided into k rectangles of width
ni and number of elements N.
2. The elements in each rectangle are rearranged to a row in the
permuted rectangle. Rectangles are taken from left to right
beginning with upper rectangles, then lower ones.
3. Inside each rectangle, the scan begins from the bottom left
corner toward upper elements.
32 Im ag e En c ry p ti o n

p1 p 2 p3 p4 p5 p6 p7 p8 p31 p23 p15 p7 p32 p24 p16 p8

p9 p10 p11 p12 p13 p14 p15 p16 p63 p55 p47 p39 p64 p56 p48 p40
p17 p18 p19 p20 p21 p22 p23 p24 p11 p3 p12 p4 p13 p5 p14 p6
p25 p26 p27 p28 p29 p30 p31 p32 p27 p19 p28 p20 p29 p21 p30 p22
p33 p34 p35 p36 p37 p38 p39 p40 p43 p35 p44 p36 p45 p37 p46 p38
p41 p42 p43 p44 p45 p46 p47 p48 p59 p51 p60 p52 p61 p53 p62 p54
p49 p50 p51 p52 p53 p54 p55 p56 p25 p17 p9 p1 p26 p18 p10 p2
p57 p58 p59 p60 p61 p62 p63 p64 p57 p49 p41 p33 p58 p50 p42 p34

(a) (b)

Figure 2.17  (a) The 8 × 8 matrix divided into rectangles. (b) The matrix after applying the 2D
Baker map.

Figure  2.17 shows an example for the chaotic map of an (8 × 8)

square image (i.e., N = 8). The secret key, Skey = [n1, n2, n3] = [2, 4, 2].
The cipher resulting from the chaotic Baker map encryption is
a permutation cipher, which does not change the histogram of the
original image. Although this cryptosystem is simple and fast to be
used for video encryption, it lacks the high degree of security, and its
computation time grows as the image size increases [51,52,61].
 3
Encryption Evaluation
Metrics

3.1 Introduction

With the application of an encryption algorithm to an image, its

pixel values change when compared with the original image. A good
encryption algorithm must make these changes in an irregular man-
ner and maximize the difference in pixel values between the original
and the encrypted images. Also, to obtain a good encrypted image, it
must be composed of totally random patterns that do not reveal any
of the features of the original image. The encrypted image has to be
independent of the original image. It should have a low correlation
with the original image [62–66].
One of the important metrics in examining an encrypted image is
the visual inspection: The more hidden the features of the image are,
the better the encryption algorithm is. Unfortunately, visual inspec-
tion only is not enough to judge the complete hiding of the contents
of the image, so other metrics are considered to evaluate the degree of
encryption quantitatively [67,68].
Diffusion is an important parameter that must be measured to judge
the encryption algorithm randomization. If an algorithm has a good
diffusion characteristic, the relation between the encrypted image and
the original image is too complex, and it cannot be predicted easily.
To measure the diffusion of any algorithm, a bit is changed in the
plainimage, and the difference between the encrypted image obtained
from the original plainimage and the encrypted image obtained from
the modified one is obtained [69].
Other tests can be used to determine some specific characteristics
of the encryption algorithms, such as the noise immunity and the
processing time.

33
34 IMAGE EN C RY P TI ON

3.2  Encryption Evaluation Metrics

This section discusses, in detail, two families of encryption ­metrics;

the first family evaluates the ability of the encryption algorithm to
­substitute the original image with an uncorrelated encrypted image.
In this family, five metrics are studied: the histogram deviation DH,
the correlation coefficient rxy, the irregular deviation DI, the h
­ istogram
uniformity, and the deviation from ideality. The second family evalu-
ates the diffusion characteristics of the encryption ­a lgorithm. In this
family, three metrics are studied: the Avalanche effect, number of
pixel change rate (NPCR), and unified average changing intensity
(UACI).

3.2.1  Histogram Deviation

The histogram deviation measures the quality of encryption in terms

of how it maximizes the deviation between the original and the
encrypted images [70]. The steps for calculating this metric are as
follows:

1. Estimate the histogram of both the original and the encrypted

images.
2. Estimate the absolute difference between both histograms.
3. Estimate the area under the absolute difference curve divided
by the total area of the image as follows:

 d 0 + d 255
∑ 
254
 + di 
2 i =1 
DH = (3.1)
M×N

where di is the amplitude of the absolute difference at the

gray level i. M and N are the dimensions of the image to be
encrypted. The higher the value of DH is, the better the qual-
ity of the encrypted image will be [70].

Although this metric gives good results about how the encrypted
image deviates from the original image, it cannot be used alone to
measure the quality of encryption as it has some limitations, explained
further in this chapter.
 En c ry p ti o n E va luati o n M e t ric s 35

3.2.2  Correlation Coefficient

A useful metric to assess the encryption quality of any image

­cryptosystem is the correlation coefficient between pixels at the same
indices in the plain- and the cipherimages [70]. This metric can be
calculated as follows:

cov( x , y )
r xy = (3.2)
D(x ) D( y )

where x and y are the plain- and cipherimages. In numerical computa-

tions, the following discrete formulas can be used:

E (x ) = 1
L ∑ x (3.3)
l =1
l

D(x ) = 1
L ∑ (x − E(x )) (3.4)
l =1
l
2

cov( x , y ) = 1
L ∑ (x − E(x ))( y − E( y )) (3.5)
l =1
l l

where L is the number of pixels involved in the calculations. The

closer the value of rxy to zero, the better the quality of the encryption
algorithm will be.

3.2.3  Irregular Deviation

The irregular deviation measures the quality of encryption in terms

of how much the deviation caused by encryption (on the encrypted
image) is irregular [71]. The steps for calculating this metric are as
follows:
1. Calculate the absolute difference between the encrypted
image and the original image.
2. Estimate the histogram H of this absolute difference matrix.
3. Estimate the mean value MH of this histogram.

www.allitebooks.com
36 IMAGE EN C RY P TI ON

4. Estimate the absolute of the histogram deviations from this

mean value as follows:
H D (i) = |H(i) − MH| (3.6)
The irregular deviation DI is calculated as follows:

∑
255
H D (i )
DI = i =0
(3.7)
M×N
The lower the value of DI, the better the encryption quality will be.

3.2.4  Histogram Uniformity

A histogram uses a bar graph to profile the occurrence of each gray

level of the image. The horizontal axis represents the gray-level value.
It begins at zero and goes to the number of gray levels. Each vertical
bar represents the number of occurrences of the corresponding gray
level in the image [72].
For image encryption algorithms, the histogram of the encrypted
image should have two properties:
1. It must be totally different from the histogram of the original
image.
2. It must have a uniform distribution, which means that the
probability of occurrence of any grayscale value is the same.

3.2.5  Deviation from Ideality

The deviation from ideality measures the quality of encryption in

terms of how the encryption algorithm minimizes the deviation of
the encrypted image from an assumed ideal encryption case. An ide-
ally encrypted image CI must have a completely uniform histogram
distribution, which means that the probability of existence of any gray
level is constant. From this definition of the ideal encrypted image
histogram, it can be formulated as

 M×N
 0 ≤ C I ≤ 255
H (C I ) =  256 (3.8)
 0 elsewhere

 En c ry p ti o n E va luati o n M e t ric s 37

The deviation from ideality can be represented as follows:

255

∑ H (C ) − H (C )
CI =0
I

D= (3.9)
M×N

where H(C) is the histogram of encrypted image. Of course, the lower

the value of D, the better the encryption quality will be.

3.2.6  Avalanche Effect

We can use the Avalanche effect metric [73,74] to test the efficiency
of the diffusion mechanism. A single bit change can be made in the
image P to give a modified image P’. Both P and P’ are encrypted to
give C and C’. The Avalanche effect metric is the percentage of differ-
ent bits between C and C’. If C and C’ differ from each other in half
of their bits, we can say that the encryption algorithm possesses good
diffusion characteristics.

3.2.7  NPCR and UACI

To test the influence of a one-pixel change on the whole image

encrypted by any encryption algorithm, two common metrics may
be used: NPCR and UACI [75]. Let the two ciphered images, whose
corresponding plainimages have only one pixel difference, be denoted
by C1 and C2 . Label the grayscale values of the pixels at grid (i,j) in
C1 and C2 by C1(i,j) and C2(i,j), respectively. Define a binary matrix
D with the same size as the images C1 and C2. Then, D(i,j) is deter-
mined from C1(i,j) and C2(i,j). If C1(i,j) = C2(i,j), then D(i,j) = 0;
otherwise, D(i,j) = 1.
The NPCR is defined as

NPCR =
∑ i, j
D(i , j )
× 100% (3.10)
M×N

The NPCR measures the percentage of different pixels in the two

images.
38 IMAGE EN C RY P TI ON

The UACI is defined as

UACI =
1 
M × N  ∑ i, j
C1 (i , j ) − C 2 (i , j ) 
255  × 100% (3.11)

It measures the average intensity of differences between the two

images. The higher the values of NPCR and UACI are, the better the
encryption will be.

3.3  Other Tests

3.3.1  Noise Immunity

The noise immunity reflects the ability of the image cryptosystem to

tolerate noise. To test the noise immunity, noise with different signal-
to-noise ratios (SNRs) is added to the encrypted image, and then the
decryption algorithm is performed. If the decrypted image is close to the
original image, we can say that the cryptosystem at hand is immune to
noise. This closeness can be verified visually or numerically with the value
of rxyd, which represents the correlation coefficient between the origi-
nal image and the decrypted image, and the peak signal-to-noise ratio
(PSNR) of the decrypted image, which is defined as follows [60,66]:
 
 
M × N × 255 2
PSNR = 10 × log10   (3.12)
M N 
∑∑ ( ( ) d ( )) 
2
 f m , n − f m , n
 
m =1 n =1

where f (m, n) is the original image, and fd (m, n) is the decrypted image.

3.3.2  The Processing Time

The processing time is the time required to encrypt and decrypt

an image. The smaller value the processing time has, the better the
encryption efficiency will be.

3.4  Testing the Evaluation Metrics

We have tested these metrics by evaluating three encrypted images with

well-known encryption quality. The first one is obtained by flipping the
Cameraman image using the Caesar cipher, which is a straightforward
 En c ry p ti o n E va luati o n M e t ric s 39

ciphering technique. The image size is 256 × 256 ­pixels. If we apply this
cipher on an image I to obtain the image J, then we will add a constant
offset to the pixel values such that the values exceeding 255 are rotated
back starting from the 0 level again. This is shown by the equation
J(m, n) = {I(m, n)+K(m, n)}mod 256, K(m, n) = 254 × I(m, n) + 255
(3.13)
where K is the cipher offset, and it is considered the key as well. We
choose the offset K of the Caesar cipher such that the location con-
taining a peak in the histogram of the original image contains a very
low value in the encrypted image, and the location containing a low
value in the original image contains a peak in the encrypted image as
shown in Figure 3.1.
Note here that the Caesar cipher is useless in image encryption,
because only a change in the image gray scale occurs keeping the
entire image features visible to the attacker, who does not know the
decryption key. It is noticed that the image was not actually encrypted.
Besides, this cipher does not possess any diffusion at all.

(a) (b)
1000
1000
900
900
800 800
700
Occurrence

700
Occurrence

600 600
500 500
400 400
300 300
200 200
100 100
0 0

0 50 100 150 200 250 0 50 100 150 200 250

Gray Level Gray Level

(c) (d)

Figure 3.1  The Cameraman image: (a) original version; (b) flipped version; (c) original image
histogram; (d) flipped image histogram.
40 IMAGE EN C RY P TI ON

(a) (b)

18000 4500
16000 4000
14000 3500
Occurrence

12000 3000

Occurrence
10000 2500
8000 2000
6000 1500
4000 1000
2000 500
0 0
0 50 100 150 200 250 0 50 100 150 200 250
Gray Level Gray Level
(c) (d)

Figure 3.2  The Coin image: (a) original version; (b) encrypted version; (c) original image
­histogram; (d) encrypted image histogram.

600

500

400
Occurrence

300

200

100

0 50 100 150 200 250

Gray Level
(a) (b)

Figure 3.3  (a) The encrypted image with the RC6 algorithm; (b) the histogram.

The second image is obtained by encrypting the Coin image, which

is a black-and-white image using the RC6 algorithm in the ECB
mode. The image size is 340 × 400 pixels. Algorithms implemented
in the ECB mode cannot encrypt a black-and-white image because of
its localized histogram. This is shown in Figure 3.2. The third image
is obtained by encrypting the Cameraman image using the RC6
­algorithm in the CBC mode as shown in Figure 3.3.
 En c ry p ti o n E va luati o n M e t ric s 41

Table 3.1  Encryption Quality Results:

Substitution
DH C.C DI D
Case 1 1.29 −1 1.26 0.98
Case 2 1.45 0.0554 1.7 1.7
Case 3 0.98 −0.0076 0.6 0.05

Table 3.2  Encryption Quality Results: Diffusion

AVALANCHE NPCR UACI
Case 1 ≅0% ≅0% ≅0%
Case 2 0.0059% 0.0118% 0.0013%
Case 3 50.07% 99.62% 16.7%

It is obvious that the first two images were encrypted badly without
any diffusion characteristics, while the last image was encrypted well,
with good diffusion characteristics. The results of each evaluation
metric for every case are tabulated in Table 3.1, and the diffusion test
results are tabulated in Table 3.2.
From these results, we can see the following for the first case:
1. The histogram deviation gives a high result of 1.29, indicating
the good performance of the Caesar cipher, which is com-
pletely a wrong decision as the entire image features are still
visible. The histogram deviation depends on the difference
between the histograms of the plainimage and the cipherim-
age, which does not necessarily mean a good encryption.
2. The correlation coefficient of −1 means that the cipherimage
is the reverse of the plainimage, which is totally true.
3. The rest of the results, including the histogram uniformity,
have judged the bad performance of the Caesar cipher correctly.
4. For the diffusion tests, the results are almost zero because the
Caesar cipher does not possess any diffusion characteristics.
For the second case,
1. The histogram deviation, again, could not judge the ­encryption
performance correctly because the plainimage has a localized
histogram.
2. For the correlation coefficient, the result indicates that the
plainimage is uncorrelated with the cipherimage, which is
42 IMAGE EN C RY P TI ON

true, but this does not mean that the plainimage is encrypted
correctly.
3. The rest of the tests have judged the encryption quality
correctly.
4. For the ECB mode, the diffusion characteristics are bad, and
this is indicated correctly with the diffusion tests.
For the third case, all test results correctly judged the good encryp-
tion and diffusion characteristics of the CBC mode.

3.5 Summary

After all these tests, we can see that

1. Visual inspection is the first test to be used. If the specifica-
tions of the plainimage were not completely hidden, we can
say that this encryption algorithm is not confident and cannot
be used to encrypt images regardless of all other tests.
2. The irregular deviation, the deviation from ideality, and the
histogram uniformity perform well for judging encryption
quality.
3. The three tests of diffusion can be used effectively to judge the
diffusion characteristics of the encryption algorithm.
 4
Homomorphic Image
Encryption

4.1 Overview

This chapter presents a new image cryptosystem. This system is based

on homomorphic image processing, which has evolved primarily as a
tool of image enhancement for images captured in bad lighting condi-
tions. The main idea of homomorphic image processing is based on
modeling the image as a product of constant illumination and vary-
ing reflectance. The product is dealt with as a summation using the
logarithmic operation. The reflectance component can be separated
using a high-pass filter, while the illumination component is sepa-
rated using a low-pass filter. Most of the image details lie in the reflec-
tance component, while the illumination component is approximately
constant [76,77].
We can carry out the encryption process in the homomorphic
domain on the reflectance component, which is the most significant
component of the image. Rather than encrypting the illumination
component, which causes redundancy in image information, it is
appended as a least-significant bit (LSB) watermark in the encrypted
reflectance component. Two algorithms are used for the encryption
of the reflectance component: the RC6 block cipher algorithm and
the chaotic Baker map scrambling algorithm. A comparison is made
between them.

4.2  Homomorphic Cryptosystem

The idea of the homomorphic cryptosystem is based on homomorphic

image processing. It is known that the image intensity can be repre-
sented as follows [78–80]:
I(m,n) = i(m,n)r(m,n) (4.1)

43
44 IMAGE EN C RY P TI ON

where i(m,n) is the light illumination, and r(m1,n2) is the reflectance

of the object to be imaged. Taking the natural logarithm of both sides
leads to
ln[I(m,n)] = ln[i(m,n)] + ln[r(m,n)] (4.2)
The illumination is approximately constant, while the reflec-
tance is variable from object to object. Thus, the term ln[i(m,n)] is
approximately constant. We can perform the encryption process on
the ln[r(m,n)] term. To avoid the redundancy resulting from the exis-
tence of two components of the image in the homomorphic domain,
we can embed the illumination component as an LSB watermark to
the encrypted reflectance component. The homomorphic cryptosys-
tem is illustrated in Figure 4.1. The 3 × 3 averaging filter shown in
Figure 4.2 is used as the low-pass filter, and the reflectance compo-
nent is obtained by subtracting the log illumination component from
the log image intensity.

Ln(i(m, n)) Watermark

LPF

I(m, n) LSB Encrypted

Ln Watermarking Image

Encryption
HPF
Algorithm

Ln(r(m, n))
(a)

Ln(i(m, n))
LSB
Encrypted + exp Decrypted
Watermark
Image Decryption Image
Extraction
Algorithm
Ln(r(m, n))
(b)

Figure 4.1  Homomorphic image cryptosystem: (a) encryption subsystem and (b) decryption
subsystem.

 1 1 1 
1  
× 1 1 1 
9
 1 1 1 

Figure 4.2  The 3 × 3 averaging filter.

 H o m o m o r p hic Im ag e En c ry p ti o n 45

The RC6 block cipher algorithm and the chaotic Baker map scram-
bling algorithm have been chosen as they are representatives of differ-
ent encryption families. The first one belongs to the family of diffusion
algorithms, and the second one belongs to the family of permutation
algorithms. The objective is to decide which family is more appropri-
ate for the homomorphic cryptosystem.

4.3  Security Analysis and Test Results

A good encryption scheme should resist all kinds of known attacks,

such as the known-plaintext attack, the ciphertext-only attack, the
statistical attack, the differential attack, and the various brute-force
attacks. The security of the homomorphic image cryptosystem was
investigated for digital images under the brute-force attack, the
­statistical attacks, and the differential attacks [81–85].
Some security analysis results, including the key space analysis, the
statistical analysis, and the differential analysis have been presented
[60,86,87]. Tests are made on the image of Lena shown in Figure 4.3.

4.3.1  Statistical Analysis

In [13], Shannon mentioned that, “It is possible to solve many kinds of

ciphers by statistical analysis.” Statistical analysis has been performed
on the homomorphic image cryptosystem, demonstrating its superior
confusion and diffusion properties, which strongly resist statistical
attacks.

3000

2500

2000
Occurrence

1500

1000

500

0
0 50 100 150 200 250
Gray Scale
(a) (b)

Figure 4.3  Lena: (a) image and (b) histogram.

www.allitebooks.com
46 IMAGE EN C RY P TI ON

4.3.1.1  Histograms of Encrypted Images  A typical example of the histo-

gram test is shown in Figures 4.4 to 4.7. From these figures, one can see
that the histogram of the encrypted image (cipherimage) by the homo-
morphic cryptosystem using the RC6 algorithm is fairly uniform and
is significantly different from that of the original image (plainimage)
as shown in Figure 4.4. This result implies to the RC6 block cipher as
shown in Figure 4.5. For the homomorphic cryptosystem using chaotic
Baker map scrambling, the result is shown in Figure 4.6. It is clear that
the histogram is different from that of the individual chaotic Baker map
encryption shown in Figure 4.7. It is known that chaotic Baker map
encryption does not change the histogram of the encrypted image from
that of the original image. On the other hand, the homomorphic cryp-
tosystem using the chaotic Baker map scrambling significantly changes
the histogram of the encrypted image.
2500

2000

1500
Distribution

1000

500

0 50 100 150 200 250

Gray Scale
(a) (b)

Figure 4.4  Encrypted image using the homomorphic cryptosystem with the RC6 algorithm: (a)
encrypted image and (b) histogram.

2500

2000
Distribution

1500

1000

500

0 50 100 150 200 250

Gray Scale
(a) (b)

Figure 4.5  Encrypted image using the RC6 algorithm: (a) encrypted image and (b) histogram.
 H o m o m o r p hic Im ag e En c ry p ti o n 47

4500
4000
3500
3000

Distribution
2500
2000
1500
1000
500
0

0 50 100 150 200 250

Gray Scale
(a) (b)

Figure 4.6  Encrypted image using the homomorphic cryptosystem with the chaotic Baker map
scrambling algorithm: (a) encrypted image and (b) histogram.

3000

2500

2000
Distribution

1500

1000

500

0 50 100 150 200 250

Gray Scale
(a) (b)

Figure 4.7  Encrypted image using the chaotic Baker map algorithm: (a) encrypted image and
(b) histogram.

4.3.1.2 Encryption Quality Measurements  The correlation coefficient

between the plain- and cipherimages, the irregular deviation, and the
deviation from ideality metrics have been used to test the quality of
each cipher algorithm. The results of these tests are shown in Table 4.1.
From the correlation test, we can see that, in all cases, the plain-
image was uncorrelated with the cipherimage, but as discussed in
Chapter 3, we cannot completely depend on this test to judge the
encryption quality correctly, as we can see that the chaotic Baker map
scrambling algorithm achieved a good result of 0.0032 although it is
only a permutation cipher.
For the irregular deviation and the deviation from ideality, we can
see that the quality of the homomorphic encryption depends mainly
48 IMAGE EN C RY P TI ON

Table 4.1  Encryption Quality Test Results for Each Cipher Algorithm

ENCRYPTION ALGORITHM Rxye DI D
Homomorphic cryptosystem with the RC6 algorithm 0.0033 0.707 0.0254
RC6 algorithm 0.0013 0.705 0.0233
Homomorphic cryptosystem with the chaotic Baker 0.0043 0.846 1.2618
map scrambling algorithm
Chaotic Baker map scrambling algorithm 0.0032 0.979 0.7155

on the core algorithm used. If the core encryption algorithm is pow-

erful, like the RC6, the homomorphic version will be powerful, and
vice versa.

4.3.2  Key Space Analysis

A good image cryptosystem algorithm should be sensitive to the

cipher keys. For the homomorphic image cryptosystem, the key space
analysis and test are summarized in the following sections.

4.3.2.1  Exhaustive Key Search  For a secure image cryptosystem, the key
space should be large enough to make the brute-force attack i­ nfeasible
[88]. The RC6 algorithm is a 128-bit encryption scheme. An exhaus-
tive key search will take 2k operations to succeed, where k is the key
size in bits. An attacker simply tries all keys, one by one, and checks
whether the given plainimage encrypts to the given cipherimage.
For a practical use of the homomorphic cryptosystem, assume that
the secret key length is 128 bits. Therefore, an opponent may try to
bypass guessing the key and directly guess all the possible combina-
tions. The opponent will need about 2128 operations to successfully
determine the key. If an opponent employs a 1000-MIPS (million
instructions per second) computer to guess the key by the brute-force
attack, the computational load is then

2128
> 10.7902831 × 10 21 years (4.3)
1000 ×10 6 × 60 × 60 × 24 × 365

This is practically infeasible.

For chaotic Baker map encryption, the key is dependent on the
width (or height) of an image. This is due to the scrambling phe-
nomena of the chaotic Baker map. For the 512 × 512 Lena image,
 H o m o m o r p hic Im ag e En c ry p ti o n 49

the number of possible keys is 10126 [52]. So, in this case, the compu-
tational load is then
10126
> 3.1710 × 10109 years (4.4)
1000 ×10 6 × 60 × 60 × 24 × 365

4.3.2.2 Key Sensitivity Test  High key sensitivity is required for

secure image cryptosystems. This means that the cipherimage can-
not be decrypted correctly if there is only a slight difference between
encryption or decryption keys [89]. This guarantees the security of
the proposed cryptosystem against brute-force attacks. Assume that
a 16-character ciphering key is used. This means that the key consists
of 128 bits.
For testing the key sensitivity of the homomorphic cryptosystem
using the RC6 algorithm, we have performed the following steps:
(a) An image is encrypted using the secret key of 32 zeroes
(in hexadecimal), and the resultant image is referred to as
encrypted image A as shown in Figure 4.8a.
(b) The same image is encrypted by making a slight modification
in the secret key (i.e., 8 and 31 zeroes [in hexadecimal]). The
change is made in the most significant digit in the secret key.
The resultant image is referred to as encrypted image B as
shown in Figure 4.8b.
(c) Again, the same image is encrypted by making another slight
modification in the secret key (i.e., 31 zeroes and 1 [in hexa-
decimal]). The change is made in the least-significant digit in

(a) (b) (c)

Figure 4.8  Key sensitivity test of the homomorphic cryptosystem with the RC6 algorithm:
(a)  encrypted image A with a key of 32 zeroes hexadecimal; (b) encrypted image B with a key of 8
and 31 zeroes hexadecimal; and (c) encrypted image C with a key of 31 zeroes and 1 hexadecimal.
50 IMAGE EN C RY P TI ON

the secret key. The resultant image is referred to as encrypted

image C as shown in Figure 4.8c.
(d) Finally, the three encrypted images A, B, and C are compared.
It is not easy to compare the encrypted images by simply observ-
ing them. So, for comparison, we calculate the correlation coefficients
between each two of the three encrypted images. Table 4.2 gives the
correlation coefficient results. It is clear from the table that no correla-
tion exists among the encrypted images even though they have been
produced using slightly different secret keys. The same results are
obtained using the RC6 algorithm, as shown in Figures  4.9a–4.9c.
The results of the correlation coefficients are tabulated in Table 4.3.
For the homomorphic cryptosystem using the chaotic Baker map
scrambling algorithm:
(a) The original image is encrypted using the secret key:
n = [10,5,12,5,10,8,14,10,5,12,5,10,8,14,10,5,12,
5,10,8,14,10,5,12,5,10,8,14,10,5,12,5,10,8,14,10,5,
12,5,10,8,14,10,5,12,5,10,8,14,10,5,12,5,10,8,14]
and the resultant image is referred to as encrypted image A
as shown in Figure 4.10a.

Table 4.2  Results of the Key Sensitivity Test for the

Homomorphic Cryptosystem with the RC6 Algorithm
IMAGE 1 IMAGE 2 Rxy
Encrypted image A Encrypted image B 0.0034
Encrypted image B Encrypted image C 0.00007
Encrypted image C Encrypted image A 0.0008

(a) (b) (c)

Figure 4.9  Key sensitivity test of RC6 algorithm: (a) encrypted image A with a key of 32 zeroes
hexadecimal: (b) encrypted image B with a key of 8 and 31 zeroes hexadecimal; and (c) encrypted
image C with a key of 31 zeroes and 1 hexadecimal.
 H o m o m o r p hic Im ag e En c ry p ti o n 51

Table 4.3  Results of the Key Sensitivity Test for

the RC6 Algorithm
IMAGE 1 IMAGE 2 Rxy
Encrypted image A Encrypted image B 0.0013
Encrypted image B Encrypted image C 0.0028
Encrypted image C Encrypted image A 0.0004

(a) (b) (c)

Figure 4.10  Key sensitivity test of the homomorphic cryptosystem with the chaotic Baker map
scrambling algorithm: (a) encrypted image A with key n; (b) encrypted image B with key n1; and
(c)  encrypted image C with key n2.

(b) The same image is encrypted by making a slight modification

in the secret key:
n1 = [5,5,5,12,5,10,8,14,10,5,12,5,10,8,14,10,5,12,
5,10,8,14,10,5,12,5,10,8,14,10,5,12,5,10,8,14,10,5,1
2,5,10,8,14,10,5,12,5,10,8,14,10,5,12,5,10,8,14].
The change is made in (10) to (5, 5) at the beginning of
the secret key. The resultant image is referred to as encrypted
image B as shown in Figure 4.10b.
(c) Again, the same image is encrypted by making another slight
modification in the secret key:
n2 = [10,5,12,5,10,8,14,10,5,12,5,10,8,14,10,5,1
2,5,10,8,14,10,5,12,5,10,8,14,10,5,12,5,10,8,14,
10,5,12,5,10,8,14,10,5,12,5,10,8,14,10,5,12,5,1
0,8,7,7]
The change is made in (14) to (7, 7) at the end of the secret
key. The resultant image is referred to as encrypted image C
as shown in Figure 4.10c.
(d) Finally, the three encrypted images A, B, and C are compared.
The correlation coefficients between each two of the three encrypted
images A, B, and C are tabulated in Table 4.4, from which it can be
52 IMAGE EN C RY P TI ON

Table 4.4  Results of the Key Sensitivity Test for

the Homomorphic Cryptosystem with the Chaotic
Baker Map Encryption Algorithm
IMAGE 1 IMAGE 2 Rxy
Encrypted image A Encrypted image B 0.9533
Encrypted image B Encrypted image C 0.8761
Encrypted image C Encrypted image A 0.9212

(a) (b) (c)

Figure 4.11  Key sensitivity test of the chaotic Baker map algorithm: (a) encrypted image A with
key n; (b) encrypted image B with key n1; and (c) encrypted image C with key n2.

said that the correlation coefficients are worse than those obtained
using the homomorphic cryptosystem with the RC6 algorithm.
Similar results are obtained using the chaotic Baker map scrambling
encryption algorithm only, as shown in Figures  4.11a–4.11c and
Table 4.5.
Another test for the key sensitivity of the homomorphic image
cryptosystem using the RC6 encryption algorithm is performed
through the following steps:
1. A 512 × 512 image is encrypted using the secret test key of 32
zeroes.
2. The encryption key is changed by changing its LSB to be 31
zeroes and 1.
3. The two ciphered images are compared.
The result is that the image encrypted with the key of 31 zeroes and
1 is totally different from the image encrypted with the all-zeros key,
although there is only 1 bit difference in the two keys. Figure 4.12a
shows the difference image between the two ciphered images. A simi-
lar test was also applied to the RC6 encryption algorithm, and the
result is shown in Figure 4.12b.
 H o m o m o r p hic Im ag e En c ry p ti o n 53

Table 4.5  Results of the Key Sensitivity Test for

the Chaotic Baker Map Encryption Algorithm
IMAGE 1 IMAGE 2 Rxy
Encrypted image A Encrypted image B 0.3247
Encrypted image B Encrypted image C 0.8762
Encrypted image C Encrypted image A 0.2877

(a) (b) (c) (d)

Figure 4.12  Difference image between the two ciphered images using (a) the homomorphic
cryptosystem with the RC6 algorithm, (b) the RC6 algorithm, (c) the homomorphic cryptosystem with
the chaotic Baker map algorithm, and (d) the chaotic Baker map scrambling algorithm.

For the homomorphic image cryptosystem using the chaotic Baker

map scrambling algorithm, the test is done using the following steps:
1. A 512 × 512 image is encrypted using the secret test key n.
2. The encryption key is changed to n1.
3. The two ciphered images are compared.
The result of this test is shown in Figure 4.12c. This test was also
applied for the chaotic Baker map scrambling only, and the result is
shown in Figure 4.12d.

4.3.3  Differential Analysis

A desirable property for the homomorphic cryptosystem is the high

sensitivity to small changes in the plainimage (single-bit change in
the plainimage) (i.e., the diffusion).
A test was performed on the 1-pixel change influence on the 256
gray-level Lena image of size 512 × 512, and the results are shown in
Table 4.6.
With respect to the NPCR and UACI estimation results in
Table  4.6, the RC6 and chaotic Baker map scrambling encryption
schemes had no sensitivity to small changes in the plainimage, but the
homomorphic cryptosystem using both schemes was highly sensitive
54 IMAGE EN C RY P TI ON

Table 4.6  NPCR and UACI Results

ALGORITHMS NPCR UACI
Homomorphic with RC6 99.605% 16.775%
RC6 0.0061% 0.0005%
Homomorphic with chaotic Baker map scrambling 100% 7.2314%
Chaotic Baker map scrambling ≅0% ≅0%

to small changes in the plainimage. Generally, these obtained results

showed that the homomorphic cryptosystem had a very powerful
­diffusion mechanism.

4.4  Effect of Noise

The test for effect of noise has been performed by adding additive white
Gaussian noise (AWGN) to the encrypted image prior to decryption.
Test results showed that the RC6 and the homomorphic algorithm
with RC6 were significantly affected by the noise in the decryption
process. In other words, these algorithms can only be used in error-
free scenarios. The experiment results are shown in Figures  4.13a
and 4.13b.
For the chaotic Baker map encryption and the homomorphic
encryption with the chaotic Baker map, the results showed that
these algorithms were more robust to noise and could work in noisy
environments. The experimental results are shown in Figures 4.13c
and 4.13d. It is clear that the chaotic Baker map decryption pro-
cess was more robust to noise than the RC6 decryption process,
and this  appears in Figure  4.14, which shows the variation of the
peak signal-to-noise ratio (PSNR) of the decrypted image with the
­signal-to-noise ratio (SNR) of the encrypted image for all algorithms.
This is attributed to the fact that the RC6 algorithm has a diffusion
mechanism in its equation f(x) = x(2x + 1)(mod 2w), which leads to
less noise immunity.

4.5 Summary

From these results, we can see that

1. In general, the encryption in the homomorphic domain can
be considered a very powerful diffusion mechanism, but this
 H o m o m o r p hic Im ag e En c ry p ti o n 55

(a) (b) (c) (d)

Figure 4.13  Decrypted images for all encryption algorithms in the presence of noise with an SNR
of 50 dB: (a) the RC6 algorithm; (b) the homomorphic cryptosystem with the RC6 algorithm; (c) the
chaotic Baker map algorithm; and (d) the homomorphic cryptosystem with the chaotic Baker map
scrambling algorithm.

60
Homomorphic encryption using RC6
Homomorphic encryption using chaotic Baker map
50 RC6
Chaotic Baker map
40
PSNR

30

20

10

0
0 5 10 15 20 25 30 35 40
SNR

Figure 4.14  The variation of the PSNR of the decrypted image with the SNR of the encrypted
image for all encryption algorithms.

comes at the expense of losses in the decrypted image due to

the LSB watermark.
2. All other parameters, like encryption quality, noise immu-
nity, and key sensitivity, depend mainly on the encryption
algorithm. For the diffusion ciphers, like RC6, the encryp-
tion quality and the key sensitivity are better than those of
the permutation ciphers, like chaotic Baker map. But, if the
transmission medium is noisy, the permutation ciphers are
preferred.

www.allitebooks.com
 5
C haotic I mag e E n cryp ti on
with D ifferent M od es
o f O per ation

5.1 Overview

Chaos theory consistently plays an active role in modern ­cryptography.

The attractiveness of using chaos as the basis for developing a cryp-
tosystem is mainly its random behavior and sensitivity to initial
conditions and parameter settings that fulfill the classic Shannon
requirements of confusion and diffusion [89]. Chaos-based algo-
rithms have shown some exceptionally good properties in many con-
cerned aspects regarding security, complexity, speed, computational
overhead, and so on. Some chaotic cryptosystems based on ergodicity
have been proposed [90,91]. A number of chaos-based image [91,92]
and random number generation algorithms [67,93–95] based on dis-
crete chaos have been proposed, but security is generally not high
enough [96,97].
This chapter discusses implementing chaotic Baker map scrambling
of image pixels using three different modes of operation: CBC, CFB,
and OFB. This implementation depends on the block size S, where
S = N × N pixels, and an IV that works as the main key. The bits of the IV
must be random and uncorrelated as much as possible to yield a powerful
encryption mechanism. In the chapter experiments, the image encrypted
is Lena, which is a 512 × 512 grayscale image, and parts of an encrypted
version of the Cameraman image are used in the IV (Figure 5.1).

5.2  Chaotic Encryption with Modes of Operation

The main objective of chaotic encryption with different modes of oper-

ation is to increase the security of chaotic encryption with m
­ oderate
computation time and adjust it to encrypt images with arbitrary
57
58 Im ag e En c ry p ti o n

Figure 5.1  Encrypted version of Cameraman image.

dimensions. We investigated chaotic encryption with three modes

of operation: CBC, CFB, and OFB [16–18]. The three modes were
tested to decide which one would increase the data-hiding ability of
the cryptosystem. Chaotic encryption with modes of operation can be
simply summarized in the following three steps:
1. Scan the image row by row.
2. Convert the scanned rows to w blocks, each with N × N pixels.
3. Encrypt these blocks using the 2D chaotic Baker map in
the CBC, CFB, or OFB mode as explained in the following
sections.
The block diagram of chaotic encryption with modes of operation
is shown in Figure 5.2.

5.3  Implementation Issues

As stated in the previous section, chaotic encryption can be imple-

mented in three different modes of operation. It is based on the 2D
chaotic Baker map as the main encryption algorithm. It is known that
the permutations induced by the Baker map behave as typical random
permutations. An IV is used as the main key. This IV must be ran-
dom to resist the brute-force attack. The XOR operations between the
bits of the IV and the bits of the data blocks change the values of the
pixels, which makes the encryption algorithm behave like a 3D Baker
map. The algorithm also uses a secondary key, which is the key used
by the Baker map to scramble the pixels.
Chaotic encryption with modes of operation is based on the seg-
mentation of the image to be encrypted into blocks. The block size is
 C h ao tic Im ag e En c ry p ti o n 59

f(1,1) f(1,N)

f(M,1) f(M,N)

P1 P2 Pw–1 Pw

P(1,n) P(1,1) P(1,n) P(1,1) P(1,n) P(1,1) P(1,n)

P (1,1)

P(n,1) P(n,n) P(n,1) P(n,n) P(n,1) P(n,n) P(n,1) P(n,n)

P1 P2 P3 Pw–1 Pw
C0 = IV
E E E E E

C1 C2 C3 Cw– Cw
(a)
C0 = IV
EK EK EK EK
P1 P2 P3 Pw–1 Pw

C1 C2 C3 Cw–1 Cw
(b)
C0 = IV
EK EK EK EK

Pw–1
P1 P2 P3 Pw

C1 C2 C3 Cw–1 Cw
(c)

Figure 5.2  Chaotic encryption implemented in (a) CBC mode, (b) CFB mode, (c) OFB mode.

an important factor that affects the performance of the cryptosystem.

The effect of the block size on the encryption quality is studied in
detail in the simulation examples section. Because the IV has similar
size to the plaintext blocks, the increase in the block size increases the
security. In addition, the chaotic encryption algorithm with modes
60 Im ag e En c ry p ti o n

of operation will be able to encrypt images with arbitrary dimensions

after segmentation into small blocks.

5.4  Simulation Examples and Discussion

Simulation experiments have been carried out to encrypt the Lena

image with the chaotic encryption system using the different modes
of operation and comparing the obtained results with the results of
the 2D chaotic Baker map encryption algorithm and the RC6 algo-
rithm. The IV has been taken as part of the encrypted Cameraman
image shown in Figure 5.1, equal in size to the selected block size.
Several block sizes have been tested:
1.
S1 = 128 × 128 pixels. The IV is a 128 × 128 pixel section of
the encrypted Cameraman image.
2.
S2 = 64 × 64 pixels.
3.
S3 = 32 × 32 pixels.
4.
S4 = 16 × 16 pixels.
5.
S5 = 8 × 8 pixels.
The encrypted versions of the Lena image are shown using the cha-
otic Baker map cryptosystem and the RC6 cryptosystem (Figure 5.3)
and the chaotic encryption with different modes of operation and
­different block sizes (Figure 5.4). It is clear from Figure 5.4 that the
performance of the chaotic encryption with different modes of opera-
tion is good except for its implementation in the OFB mode with
large block sizes.
To check the noise immunity of each of the cryptosystems men-
tioned in this chapter, additive white Gaussian noise (AWGN) with

Figure 5.3  Encrypted images using traditional cryptosystems: (a) encrypted image using the
chaotic Baker map cryptosystem; (b) encrypted image using the RC6 cryptosystem.
 C h ao tic Im ag e En c ry p ti o n 61

CBC CFB OFB

S1

S2

S3

S4

S5

Figure 5.4  Encrypted images using chaotic encryption with different modes of operation and
different block sizes.

a signal-to-noise ratio (SNR) of 5 dB was added to the encrypted

images, and then the decryption was performed. The decryption
results of the chaotic Baker map cryptosystem and the RC6 crypto-
system (Figure 5.5) and the chaotic encryption with different modes
of operation (Figure 5.6) are shown. From visual inspection, it is clear
that the chaotic Baker map cryptosystem has the highest noise immu-
nity. It is also clear that the chaotic encryption with different modes
of operation is more immune to noise than the RC6 cryptosystem.
Table 5.1 shows the values of the evaluation metrics for all cryp-
tosystems mentioned in this chapter; rxye represents the correlation
62 Im ag e En c ry p ti o n

Figure 5.5  Decrypted images using the traditional cryptosystems at SNR = 5 dB: (a) decrypted
image using the chaotic Baker map cryptosystem; (b) decrypted image using the RC6 cryptosystem.

CBC CFB OFB

S1

S2

S3

S4

S5

Figure 5.6  Decrypted images using the chaotic algorithm with different modes of operation and
different block sizes at SNR = 5 dB.
Table 5.1  Numerical Evaluation Metrics for All Cryptosystems
CHAOTIC ENCRYPTION WITH MODES OF OPERATION
CBC CFB OFB
TEST S1 S2 S3 S4 S5 S1 S2 S3 S4 S5 S1 S2 S3 S4 S5 ENCRYPTION RC6
D 0.71 0.72 0.71 0.71 0.72 0.71 0.71 0.71 0.72 0.71 0.71 0.71 0.7 0.67 0.66 0 0.71
rxye 1.4 × 4.7 × 1.3 × 1.7 × −2 × 3.7 × 1.5 × 2× −6.5 × 7.4 × −7.1 × 2.5 × 2.3 × 13.5 × 41.8 × 10−4 1.3 ×
10−3 10−6 10−3 10−3 10−3 10−3 10−3 10−3 10−3 10−4 10−3 10−4 10−3 10−3 10−3 10−3
DI 0.71 0.7 0.71 0.71 0.7 0.71 0.71 0.71 0.7 0.71 0.7 0.71 0.73 0.74 0.95 0.98 0.71
rxyd 0.57 0.57 0.57 0.57 0.57 0.57 0.57 0.57 0.57 0.56 0.7 0.7 0.7 0.7 0.69 0.87 2.5 ×
10−3
Note: Decryption was performed in the presence of an AWGN with SNR = 5 dB.
C h ao tic Im ag e En c ry p ti o n
63
64 Im ag e En c ry p ti o n

coefficient between the original and encrypted image, and rxyd

represents the correlation coefficient between the original and
decrypted image affected by 5-dB AWGN. From the results obtained
and shown in this table, it is clear that the chaotic Baker map cryp-
tosystem is the most immune to noise, but it is less secure than the
RC6 cryptosystem and the chaotic encryption with modes of opera-
tion. It is also clear that the RC6 cryptosystem is very sensitive to the
presence of noise. We can conclude that the chaotic encryption with
different modes of operation achieves the trade-off between the level
of security and the noise immunity.
The effect of noise on the chaotic encryption with different modes of
operation and block sizes was also studied, and the results are given in
Figure 5.7. This figure shows the variation of the peak signal-to-noise
ratio (PSNR) of the decrypted image with the SNR of the encrypted
image. It is clear from this figure that all modes of operation have
approximately the same performance in the presence of noise, and the
performance is improved at high SNR values. It is also clear that the
block size has no effect on the noise immunity of the algorithm in all
implementation modes.
The histograms of the original Lena image and the encrypted
images with all cryptosystems mentioned in this chapter are shown in
Figures 5.8 and 5.9, respectively. It is clear that the histogram unifor-
mity is not achieved with the chaotic Baker map cryptosystem, which
is a weakness of this cryptosystem. The RC6 cryptosystem achieves
histogram uniformity. The chaotic encryption with different modes
of operation also achieves histogram uniformity for all implementa-
tions except the implementations in the OFB mode with large block
sizes.
It is known that the RC6 cryptosystem has long processing time
as compared to the chaotic Baker map cryptosystem. As a result,
our study for the processing time was restricted to the comparison
between the processing time of the chaotic encryption with differ-
ent modes of operation and the chaotic Baker map cryptosystem.
Figure 5.10 shows the variation of the normalized processing time of
the chaotic encryption algorithm implemented in the different modes
of operation with the block length. Normalization was performed by
dividing the processing time of the chaotic encryption algorithm with
different modes of operation by the processing time of the chaotic
 C h ao tic Im ag e En c ry p ti o n 65

34
32
30 S1
S2
28 S3
26 S4
PSNR S5
24
22
20
18
16
14
0 5 10 15 20 25 30 35 40
SNR
(a)
34
32
S1
30 S2
28 S3
S4
26 S5
PSNR

24
22
20
18
16
14
0 5 10 15 20 25 30 35 40
SNR
(b)
40
S1
35 S2
S3
S4
30 S5
PSNR

25

20

15
0 5 10 15 20 25 30 35 40
SNR
(c)

Figure 5.7  Variation of the PSNR of the decrypted image with the SNR of the encrypted image
in the chaotic encryption with different modes of operation for (a) the CBC mode, (b) the CFB mode,
(c) the OFB mode.

Baker map cryptosystem. The processing time includes the ­encryption

and decryption times. It is clear from Figure 5.10 that the process-
ing times of all implementations are approximately the same for each
block length. It is also clear that large block sizes lead to shorter pro-
cessing times because the encryption and decryption are performed
for fewer blocks.
66 Im ag e En c ry p ti o n

3000 3000

2500 2500
Occurrence

Occurrence
2000 2000

1500 1500

1000 1000

500 500

0 0

0 50 100 150 200 250 0 50 100 150 200 250

Gray Level Gray Level
(a) Histogram of the original image. (b) Histogram of the encrypted image using the
chaotic Baker map cryptosystem.

2500

2000
Occurrence

1500

1000

500

0 50 100 150 200 250

Gray Level
(c) Histogram of the encrypted image using the
RC6 cryptosystem.

Figure 5.8  Histograms of the original image and the encrypted images with the traditional
cryptosystems.
 C h ao tic Im ag e En c ry p ti o n 67

CBC CFB OFB

2500 2500 2500

2000 2000 2000

Occurrence

Occurrence

Occurrence
1500 1500 1500

1000 1000 1000

500 500 500

0 0 0
0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250
Gray Level Gray Level Gray Level
S1
2500 2500 2500

2000 2000 2000

Occurrence

Occurrence

Occurrence
1500 1500 1500

1000 1000 1000

500 500 500

0 0 0
0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250
Gray Level Gray Level Gray Level

S2
2500 2500 2500

2000 2000 2000

Occurrence

Occurrence

Occurrence
1500 1500 1500

1000 1000 1000

500 500 500

0 0 0
0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250
Gray Level Gray Level Gray Level
S3
2500 2500 2500

2000 2000 2000

Occurrence

Occurrence

Occurrence

1500 1500 1500

1000 1000 1000

500 500 500

0 0 0
0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250
Gray Level Gray Level Gray Level
S4
2500 2500 2500

2000 2000 2000

Occurrence

Occurrence

Occurrence

1500 1500 1500

1000 1000 1000

500 500 500

0 0 0
0 50 100 150 200 250 0 50 100 150 200 250 0 50 100 150 200 250
Gray Level Gray Level Gray Level
S5

Figure 5.9  Histograms of the encrypted images using chaotic encryption with different modes
of operation and different block sizes.
68 Im ag e En c ry p ti o n

Normalized Processing Time

5

4 S5

2
S4
1
S3 S2 S1
0
8 16 32 64 128
Block Size
(a)
6
Normalized Processing Time

4 S5

1 S4
S3 S2 S1
0
8 16 32 64 128
Block Size
(b)
6
Normalized Processing Time

4 S5

2
S4
1
S3 S2 S1
0
8 16 32 64 128
Block Size
(c)

Figure 5.10  The normalized processing time for the chaotic image encryption with different
modes and block sizes: (a) CBC mode; (b) CFB mode; (c) OFB mode.

5.5 Summary

1. Chaotic image encryption with different modes of operation

can be used to adjust the 2D Baker map to encrypt images
with arbitrary dimensions.
2. Chaotic encryption with the CBC mode is the best of all
implementations tested in this chapter.
 C h ao tic Im ag e En c ry p ti o n 69

3. Generally, the encryption quality of all modes increases with

the decrease of the block size, except for the OFB mode,
especially with S = S3, S4, and S5.
4. The chaotic Baker map has the worst results in all tests except
for the noise immunity, and the RC6 algorithm has interme-
diate results.
5. The processing time increases with the decrease in the block
size.
 6
D iffusi on M echanism
for D ata E n cryp ti on
in the ECB M od e

6.1 Introduction

The implementation of any block cipher algorithm depends on the

mode of operation, which governs the relation between blocks d ­ uring
the encryption process. The ECB mode is one of the possible modes
of operation. In this mode, each block is encrypted independently.
Unlike different modes of operation, this mode allows parallel pro-
cessing, which is a great advantage. Unfortunately, because the
encryption of each block does not depend on other blocks, an adver-
sary can replace any block with a previously intercepted block without
detection; hence, the message is hacked without the need to know
the key. This major security problem is called the block independency
problem. In addition, identical plaintext blocks are encrypted to iden-
tical ciphertext blocks, so symmetrical large data patterns, like those
in images, cannot be hidden using any encryption algorithm imple-
mented in the ECB mode [98–102].
Two actions are discussed in this chapter to solve the block inde-
pendency problem in the ECB mode [103]. The first action is to make
the values of the data bytes functions of their positions in the data
stream. The second action is to perform a diffusion process between
these bytes. These two actions are preprocessing actions that are per-
formed prior to the application of any block cipher algorithm imple-
mented in the ECB mode. First, an addition step, which is used to
remove any identical plaintext blocks, is performed. Then, an sub-
stitution permutation network (SPN) is used to make the diffusion
between the bytes. A schematic diagram of this mechanism is shown
in Figure 6.1.

71
72 Im ag e En c ry p ti o n

Plaintext Preprocessing Encryption in the ECB

Ciphertext
network mode

(a)

Plaintext Reverse of pre- Decryption in the ECB

Ciphertext
processing network mode

(b)

Figure 6.1  The suggested mechanism for encryption and decryption in the ECB mode:
(a) ­encryption; (b) decryption.

6.2  The Preprocessing Network

The disadvantages of the ECB mode limit its applicability in modern

cryptography despite its advantage of parallel processing [104]. If we
can preprocess the data to diffuse bytes together before encryption,
these disadvantages will be avoided. The preprocessing network con-
sists of two parts: an addition part and an SPN.

6.2.1  The Addition Part

The objective of the addition process is to make the byte values func-
tions of their positions in the data stream. This process helps in elimi-
nating identical plaintext blocks. A function of the byte position is
added to the byte value as follows:

B(i) = (B(i) + f(i)) mod 256 (6.1)

where i is the position of the byte in the data stream, and B is the byte
value.
The function f(i) is used because it satisfies the following conditions:

Randomness. The resulting values from this function for every

i are random with equal probability.
Irregularity. The differences between each byte and its neigh-
bors in the data stream after the addition of this function to
the byte values are unpredictable and unrepeatable.
Nonperiodicity. This function is nonperiodic because it satisfies
the following condition:

f(i) ≠ f(i + n) ≠ f(i + 2n) ≠ f(i + 3n) ≠ … (6.2)

 Dif f usi o n M ec h a nism f o r Data En c ry p ti o n 73

Here, we test three functions to find the one that satisfies the
given conditions:

f 1 (i) = 7 × (i + 13) mod 256 (6.3)

f 2 (i) = (7 × (i + 3))5 mod 256 (6.4)

f3 (i ) = fix ( )
7 × (i + 13)5 mod 256 (6.5)

where the fix function is used to round the value of its argument
toward the lower nearest integer.
The functions have been tested for values from 1 to 1024, and the
results are given in Figure  6.2. The probability density functions
(PDFs) of the outputs of these functions are shown in Figure 6.3.
From these figures, we can see that f 3(i) gives a random, irregular,
and nonperiodic output with a uniform PDF.

6.2.2  The SPN

The SPN is used to diffuse the bytes of the data together after the
addition step. First, the data are divided into blocks of n bits. This
network block size n is different from the encryption algorithm block
size w. After that, a chain of XOR operations is performed as shown
in Figure 6.4. The subkey K1 works as an initial key for the first XOR,
and then the result is XORed with the next block up to the end of the
plaintext. Each block resulting from this chain of XORs is permuted
bit-by-bit as shown in Figure 6.5. A block-based permutation is per-
formed after that. Finally, another chain of XORs is implemented
beginning with a subkey K 2.

6.3  Implementation Issues

The preprocessing network performance depends mainly on

the ­correct determination of the SPN block size n and its rela-
tion to the  ­encryption  algorithm block size w. The effect of the
­normalized block size (n/w) on the diffusion and block dependency
is studied.
74 Im ag e En c ry p ti o n

300
250
200
150
f(i)

100
50
0
0 200 400 600 800 1000 1200
i
(a)
300

250

200

150
f(i)

100

50

0
0 200 400 600 800 1000 1200
i
(b)

300
250
200
150
f(i)

100
50
0
0 200 400 600 800 1000 1200
i
(c)

Figure 6.2  Variation of (a) f1(i), (b) f2(i), (c) f3(i) with i.

6.3.1  Effect of the Normalized Block Size on Diffusion

Diffusion is simply hiding the relation between the plaintext and the
ciphertext. If a small change in the plaintext (one bit) makes a large
change in the ciphertext (half of its bits), then the Avalanche effect
is evident, and the algorithm has a powerful diffusion mechanism
[105–110]. This is guaranteed if we can ensure that a bit change in
 Dif f usi o n M ec h a nism f o r Data En c ry p ti o n 75

× 10–3
4

3
Probability

0
0 50 100 150 200 250 300
f(i)
(a)
0.05

0.04
Probability

0.03

0.02

0.01

0
0 50 100 150 200 250 300
f(i)
(b)
× 10–3
6
5
4
Probability

3
2
1
0
0 50 100 150 200 250 300
f(i)
(c)

Figure 6.3  PDFs of the outputs of (a) f1(i), (b) f2(i), (c) f3(i).
76 Im ag e En c ry p ti o n

S1 S2 Si SL

K1

Q1 Q2 Qi QL
Bit permutation Bit permutation Bit permutation Bit permutation

S´L S´L–1 S´L–i+1 S´1

K2

Q´1 Q´2 Q´i Q´L

Figure 6.4  Operation of the SPN.

Qi

b1 b2 bh–1 bh

bh bh–1 b2 b1

S´i

Figure 6.5  The bit permutation in the SPN.

the plaintext will affect at least one bit in each block of the input to the
encryption algorithm, and the diffusion mechanism in the encryption
algorithm will propagate this change through each block.
From Figure 6.6, we can see that if a change takes place in an SPN
block (Si), all the network blocks from Q'L-i+1 to Q'L will be affected,
where L is the number of network blocks. The output odd-number
 Dif f usi o n M ec h a nism f o r Data En c ry p ti o n 77

S1 S2 Si SL–1 SL

K1

Change free
block
Block with
change

K2

Q´1 Q´2 Q´L–i+1 Q´L–1 Q´L

To the encryption algorithm

Figure 6.6  Diffusion effect of the SPN on the data blocks before encryption.

blocks from Q'1 to Q'i will also be affected. The even-number blocks
up to Q'i will not be affected due to performing XORs between the
same patterns.
To ensure the propagation of changes in each block of the encryp-
tion algorithm, a proper choice of the SPN block size n is half the
encryption algorithm block size w, (n/w = 0.5). Because w in the cases
of the Advanced Encryption Standard (AES) and the RC6 algorithms
is 128 bits, the network block size used should be 64 bits.
This network has the ability to propagate the changes made in the
addition part and diffuse the bytes of the data together.

6.3.2  Effect of the Normalized Block Size on Block Dependency

The block independency problem gives the opportunity for a hacker to

exchange any cipher block without detection. The preprocessing net-
work solves this problem by making this exchange affect other blocks
besides the replaced block in the decryption process.
On the other hand, with the SPN, if a block is received in error,
this error will propagate through other blocks during the decryption
process, and these blocks will be decrypted incorrectly even if they
78 Im ag e En c ry p ti o n

From the decryption algorithm

P´1 P´i–1 P´i P´i+1 P´m

K2

K1

P1 Pi–1 Pi Pi+1 Pm

Figure 6.7  The error propagation due to the inversion of the SPN effect in the decryption process.

were received correctly. So, there is a need for a trade-off between the
block dependency and the error propagation.
If an encrypted block was received in error, the application of the
decryption algorithm on this block will propagate the error in two
neighboring network blocks in different manners. This is attributed
to the randomization of the diffusion mechanism in the decryption
algorithm. If an error occurs in an intermediate encrypted block, the
error spreads over three decrypted blocks. If this error occurs in the
first or the last encrypted block, the error spreads over two decrypted
blocks only. This is shown in Figure  6.7, where the shaded blocks
refer to network blocks with different errors, and white blocks refer to
the absence of errors. Every decrypted block of size w is represented
by two network blocks. From this figure, we can see that the block
dependency is ensured, but the error propagation slightly increases
compared to the CBC mode.

6.4  Simulation Examples

Several experiments have been performed to test the effect of the sug-
gested mechanism on image encryption in the ECB mode. Both the
AES and RC6 algorithms have been used as the encryption algo-
rithms. The Cameraman image shown in Figure 6.8 has been used in
these experiments.
 Dif f usi o n M ec h a nism f o r Data En c ry p ti o n 79

1000
900
800
700

Occurrence
600
500
400
300
200
100

0 50 100 150 200 250

(a) Gray Scale
(b)

Figure 6.8  (a) The Cameraman image; (b) the histogram.

6.4.1  Encryption Quality

One of the important factors in examining the encrypted image is

visual inspection. The Cameraman image has been encrypted with
the AES and the RC6 algorithms in the ECB, CBC, CFB, and OFB
modes and the ECB mode with preprocessing by the proposed net-
work for comparison purposes. The results of these experiments are
shown in Figures  6.9 and 6.10, respectively. From these obtained
results, it is clear that the encryption in the ECB mode failed to hide
the details at the upper slice of the Cameraman image because this
slice is flat in intensity. The encryption in the ECB mode with the
proposed network solves the problem.
The histograms of all images in Figures 6.9 and 6.10 are shown in
Figures 6.11 and 6.12, respectively. The histogram uniformity in all
figures ensures the success of all encryption algorithms to achieve the
required randomness. The values of the encryption quality metrics DI
and D for the encrypted images in Figures 6.9 and 6.10 are tabulated
in Table 6.1. These results show that all modes achieve the random-
ness of the data.

6.4.2 Diffusion

The diffusion metrics NPCR, UACI, and the Avalanche effect metric
have been evaluated for three cases: changing a single bit in the first
pixel, changing a single bit in the midpixel, and changing a single
80 Im ag e En c ry p ti o n

(a) (b) (c)

(d) (e)

Figure 6.9  The Cameraman image encrypted with the AES in (a) ECB mode with preprocessing,
(b) ECB mode, (c) CBC mode, (d) CFB mode, (e) OFB mode.

(a) (b) (c)

(d) (e)

Figure 6.10  The Cameraman image encrypted with the RC6 algorithm in (a) ECB mode with
preprocessing, (b) ECB mode, (c) CBC mode, (d) CFB mode, (e) OFB mode.
 Dif f usi o n M ec h a nism f o r Data En c ry p ti o n 81

600

500

400

Occurrence
300

200

100

0 50 100 150 200 250

Gray Scale
(a)

600 600

500 500

400 400
Occurrence

300 Occurrence 300

200 200

100 100

0 0

0 50 100 150 200 250 0 50 100 150 200 250

Gray Scale Gray Scale
(b) (c)

600 600

500 500

400
Occurrence

400
Occurrence

300 300

200 200

100 100

0 0
0 50 100 150 200 250 0 50 100 150 200 250
Gray Scale Gray Scale
(d) (e)

Figure 6.11  The histograms of the encrypted images in Figure 6.9.

bit in the last pixel. The results of the diffusion tests are shown in
Table 6.2. From this table, we can see that the diffusion between the
blocks of data depends mainly on the mode of operation, not on the
encryption algorithm. The ECB and OFB modes do not possess any
diffusion characteristics. The ECB mode with the preprocessing net-
work gives the best results, unlike the CBC and CFB modes, which
has low values in the Avalanche effect metric, with the changes occur-
ring close to the end of the data.
82 Im ag e En c ry p ti o n

600

500

400

Occurrence
300

200

100

0 50 100 150 200 250

Gray Scale

(a)

600
600

500 500

400 400
Occurrence

Occurrence

300 300

200 200

100 100

0 0

0 50 100 150 200 250 0 50 100 150 200 250

Gray Scale Gray Scale

(b) (c)

600 600

500 500

400 400
Occurrence

Occurrence

300 300

200 200

100 100

0 0

0 50 100 150 200 250 0 50 100 150 200 250

Gray Scale Gray Scale

(d) (e)

Figure 6.12  The histograms of the encrypted images in Figure 6.10.

6.4.3  Encryption of Images with Few Details

An image with few details has a large amount of adjacent pixels with
similar values. Examples of these images are the medical image in
Figure 6.13 and the logo image in Figure 6.14.
The encrypted versions of the medical image and the logo image
with both the AES and the RC6 algorithm in the ECB mode with
and without the preprocessing network are shown in Figures  6.15
Table 6.1  Encryption Quality Results
RC6 AES
METRIC CBC CFB OFB ECB ECB WITH PREPROCESSING CBC CFB OFB ECB ECB WITH PREPROCESSING
DI 0.602 0.596 0.597 0.597 0.598 0.601 0.603 0.603 0.606 0.604
D 0.0513 0.0486 0.0503 0.0452 0.046 0.0485 0.0458 0.0472 0.052 0.048
Dif f usi o n M ec h a nism f o r Data En c ry p ti o n
83
 84

Table 6.2  Diffusion Test Results

RC6 AES
CBC CFB OFB ECB ECB WITH CFB OFB ECB ECB WITH
TEST THE CHANGED BIT (%) (%) (%) (%) PREPROCESSING (%) CBC (%) (%) (%) (%) PREPROCESSING (%)
NPCR Bit in the first pixel 99.61 99.58 ≅0 0.02 99.6 99.65 99.6 ≅0 0.02 99.6
Bit in the middle pixel 50.03 50 ≅0 0.02 99.65 50.01 50.03 ≅0 0.02 99.56
Bit in the last pixel 0.02 ≅0 ≅0 0.02 99.6 0.024 ≅0 ≅0 0.02 99.58
UACI Bit in the first pixel 16.61 16.7 ≅0 0.01 16.66 16.76 16.62 ≅0 0.005 16.88
Bit in the middle pixel 8.4 8.4 ≅0 0.01 16.79 8.37 8.3 ≅0 0.005 16.95
Bit in the last pixel 0.005 ≅0 ≅0 0.01 16.8 0.005 ≅0 ≅0 0.003 16.92
Im ag e En c ry p ti o n

Avalanche Bit in the first pixel 50.07 49.96 ≅0 0.01 50.03 50.08 49.99 ≅0 0.011 50.03
effect metric Bit in the middle pixel 25.11 25.04 ≅0 0.01 50.11 25.09 25.15 ≅0 0.012 50.08
Bit in the last pixel 0.01 ≅0 ≅0 0.01 50.04 0.014 ≅0 ≅0 0.011 50.08
 Dif f usi o n M ec h a nism f o r Data En c ry p ti o n 85

Figure 6.13  Medical image.

Figure 6.14  Logo image.

(a) (b)

(c) (d)

Figure 6.15  The medical image encrypted with (a) the RC6 algorithm in the ECB mode, (b) the
AES in the ECB mode, (c) the RC6 algorithm in the ECB mode with preprocessing, (d) the AES in the
ECB mode with preprocessing.

and 6.16, respectively. The histograms of these images are shown in

Figures  6.17 and 6.18, respectively. From the obtained results, we
notice that the encryption in the ECB mode fails with these images.
We notice also that the preprocessing network enhances dramatically
the encryption quality of these encrypted images with the ECB mode.
86 Im ag e En c ry p ti o n

(a) (b)

(c) (d)

Figure 6.16  The logo image encrypted with (a) the RC6 algorithm in the ECB mode, (b) the AES
in the ECB mode, (c) the RC6 algorithm in the ECB mode with preprocessing, (d) the AES in the ECB
mode with preprocessing.
 Dif f usi o n M ec h a nism f o r Data En c ry p ti o n 87

×104
4
3.5
3

Occurrence
2.5
2
1.5
1
0.5
0
0 50 100 150 200 250
Gray Scale
(a)

9000 10000
9000
8000
8000
7000
7000
Occurrence

Occurrence
6000
6000
5000 5000
4000 4000
3000 3000
2000 2000
1000 1000
0 Gr
aySc
ale 0
0 50 100 150 200 250 0 50 100 150 200 250
Gray Scale Gray Scale
(b) (c)
2500
2500
2000
2000
Occurrence
Occurrence

1500
1500

1000 1000

500 500

0 0
0 50 100 150 200 250 0 50 100 150 200 250
Gray Scale Gray Scale
(d) (e)

Figure 6.17  The histograms of the medical image: (a) original, (b) encrypted with the RC6 algo-
rithm, (c) encrypted with the AES, (d) encrypted with the RC6 algorithm in the ECB mode with
preprocessing, (e) encrypted in AES in the ECB mode with preprocessing.
88 Im ag e En c ry p ti o n

8000
7000
6000

Occurrence
5000
4000
3000
2000
1000
0
0 50 100 150 200 250
Gray Scale
(a)
1800
1600 1600
1400 1400
1200 1200
Occurrence

Occurrence
1000 1000
800 800
600 600
400 400
200 200
0 0
0 50 100 150 200 250 0 50 100 150 200 250
Gray Scale Gray Scale
(b) (c)
600
600
500
500
Occurrence

400
Occurrence

400
300
300
200 200

100 100

0 0
0 50 100 150 200 250 0 50 100 150 200 250
Gray Scale Gray Scale
(d) (e)

Figure 6.18  The histograms of the logo image: (a) original, (b) encrypted with the RC6 algorithm,
(c) encrypted with the AES, (d) encrypted with the RC6 algorithm in the ECB mode with preprocess-
ing, (e) encrypted in AES in the ECB mode with preprocessing.

6.5 Summary

In this chapter, an efficient approach has been presented to solve the

problems associated with encryption in the ECB mode. A preprocess-
ing network processes the data before encryption to diffuse the data
bytes together prior to the application of the encryption algorithm in
the ECB mode. The test results have shown that any encryption algo-
rithm implemented in the ECB mode with the preprocessing network
achieves good diffusion characteristics and high encryption quality
without losing the parallel processing advantage.
 7
O rtho g onal Frequen cy
D i v isi on M ultiple xin g

7.1 Introduction

The main objective of this book is to study image encryption from a

communication perspective. So, we need to determine if an encryp-
tion algorithm fits the communication requirements. It is time now to
switch to wireless communication to understand the ­communication
process and its limitations and hence to study the performance of the
decryption algorithms after the communication process. We con-
sider multicarrier modulation (MCM), especially orthogonal fre-
quency division multiplexing (OFDM), in our study because it is the
new trend in wireless communication systems. MCM is used not
only in the physical layer of several wireless network standards such
as Institute of Electrical and Electronics Engineers (IEEE) 802.11a,
IEEE 802.16a, and HIPERLAN2 (High-Performance Radio Local
Area Network Type 2), but also in HDTV (high-definition television)
applications that include image communication [111,112]. OFDM
overcomes the effects of multipath fading by breaking the signal into
several narrow-bandwidth carriers. This results in a low symbol rate
reducing the amount of ISI (intersymbol interference). The high toler-
ance to multipath fading makes OFDM more suited to transmissions
with a high data rate in terrestrial environments compared to single-
carrier transmissions.
The transmission frequency, receiver velocity, and required mul-
tipath tolerance all determine the most suitable transmission mode to
use. Doppler spread is caused by rapid changes in the channel response
due to movement of the receiver through a multipath environment. It
results in random frequency modulation of the OFDM ­subcarriers,
leading to signal degradation. The amount of Doppler spread is pro-
portional to the transmission frequency and the velocity of movement.

89
90 Im ag e En c ry p ti o n

0.8

0.6

0.4
TX Power

0.2

–0.2

–0.4
–10 –8 –6 –4 –2 0 2 4 6 8 10
Frequency Carrier Spacing

Figure 7.1  Frequency response of the subcarriers in a seven-tone OFDM signal.

The closer the subcarriers are spaced, the more ­susceptible the OFDM
signal is to Doppler spread, so the different transmission modes in dig-
ital audio broadcasting (DAB) and HDTV allow a trade-off between
the amount of multipath protection (length of the guard period) and
the Doppler spread tolerance [113].
The OFDM signal has the spectrum shown in Figure 7.1. In the
frequency domain, each OFDM subcarrier has several sinc-shaped
frequency responses. The sinc shape has a narrow main lobe with many
side lobes that decay slowly with the magnitude of the frequency shift
away from the center. Each carrier has a peak at the center frequency
and nulls evenly spaced with a frequency gap equal to the carrier spac-
ing [114,115].

7.2  Basic Principles of OFDM

All wireless communication systems use a modulation scheme to map

the information signal to a form that can be effectively transmitted
over the communication channel. A wide range of modulation schemes
has been developed, with the most suitable one depending on whether
the information signal is an analog waveform or a digital signal.
 O r t h o g o n a l F req uen cy Di v isi o n Multip l e x in g 91

Common analog modulation schemes include f­requency m ­ odulation

(FM), amplitude modulation (AM), and phase modulation (PM).
Common single-carrier modulation schemes for digital communica-
tions include amplitude shift keying (ASK), frequency shift keying
(FSK), phase shift keying (PSK), and quadrature amplitude modula-
tion (QAM) [115].
OFDM is different from frequency division multiplexing (FDM)
in several ways. In conventional broadcasting, each radio station
transmits on a different frequency, effectively using FDM to main-
tain a separation between the stations. There is, however, no coordi-
nation  or synchronization between each of these stations. With an
OFDM transmission system such as DAB or digital video broad-
casting (DVB), the information signals from multiple stations are
combined into a single multiplexed stream of data. These data are
then transmitted using an OFDM ensemble that is made from dense
packing of several subcarriers. All the subcarriers within the OFDM
signal are time and frequency synchronized to each other, allowing
the interference between subcarriers to be carefully controlled. These
multiple subcarriers overlap in the frequency domain but cause small
effects of intercarrier interference (ICI) due to the orthogonal nature
of the modulation. Typically, with FDM the transmission signals need
to have a large frequency guard band between channels to prevent
interference. This lowers the overall spectral efficiency. However, with
OFDM, the orthogonal packing of the subcarriers greatly reduces
this guard band, improving the spectral efficiency [116–122].
Each of the carriers in an FDM transmission system can use an
analog or digital modulation scheme. There is no synchronization
between the transmissions, so one station could transmit using FM
and another in a digital form using FSK. In a single OFDM trans-
mission system, all the subcarriers are synchronized to each other,
restricting the transmission to digital modulation schemes. OFDM
is symbol based and can be thought of as a large number of low-bit-
rate carriers transmitted in parallel. All these carriers transmit using
synchronized time and frequency, forming a single block of spectrum
to ensure that the orthogonal nature of the structure is maintained.
Because these multiple carriers form a single OFDM transmission,
they are commonly referred to as subcarriers, with the term carrier
reserved for describing the radio-frequency (RF) carrier mixing the
92 Im ag e En c ry p ti o n

signal from baseband. There are several ways of looking at what makes
the subcarriers in an OFDM signal orthogonal and why this prevents
interference between them [116–119].

7.2.1 Orthogonality

Signals are orthogonal if they are mutually independent of each other.

Orthogonality is a property that allows multiple information signals to
be transmitted perfectly over a common channel and detected without
interference. Loss of orthogonality results in blurring between these
information signals and degradation in communications.
Many common multiplexing schemes are inherently orthogonal.
Time division multiplexing (TDM) allows transmission of multiple
information signals over a single channel by assigning unique time slots
to each separate information signal. During each time slot, only the
signal from a single source is transmitted, preventing any i­ nterference
between the multiple information sources. Because of this, TDM is
orthogonal in nature [116–119]. In the frequency domain, most FDM
systems are orthogonal as each of the separate transmission signals
is well spaced out in frequency, preventing interference. The term
OFDM has been reserved for a special form of FDM. The subcarri-
ers in an OFDM signal are spaced as close as is theoretically possible
while maintaining orthogonality between them.
OFDM achieves orthogonality in the frequency domain by allo-
cating each of the separate information signals onto different subcar-
riers. OFDM signals are made up from a sum of sinusoids, with each
one corresponding to a subcarrier. The baseband frequency of each
subcarrier is chosen to be an integer multiple of the inverse of the
symbol time, resulting in all subcarriers having an integer number of
cycles per symbol. As a consequence, the subcarriers are orthogonal
to each other [123–130].
Sets of functions are orthogonal to each other if they match the
conditions in Equation (7.1). If any two different functions within
the set are multiplied and integrated over a symbol period, the result
is zero for orthogonal functions. If we look at a matched receiver for
one of the orthogonal functions, then the receiver will only see the
result for that function. The results from all other functions in the set
integrate to zero and thus have no effect [115].
 O r t h o g o n a l F req uen cy Di v isi o n Multip l e x in g 93

∫ s (t ) s (t ) dt = {
0
i j
c
0
i= j
i≠ j (7.1)

Equation (7.2) shows a set of orthogonal sinusoids that represent

the subcarriers for an unmodulated real OFDM signal [115,116].

 sin(2πkf 0 t ) 0 < t <T k = 1, 2,....N

s k (t ) =  (7.2)
 0 otherwise

where f 0 is the subcarrier spacing, N is the number of subcarriers, and

T is the symbol period. Because the highest-frequency component is
Nf 0, the transmission bandwidth is also Nf 0.

7.2.2  Frequency Domain Orthogonality

In the frequency domain, each OFDM subcarrier has a sinc sin(x)/x

frequency response. This is a result of the symbol time corresponding
to the inverse of the carrier spacing. As far as the receiver is con-
cerned, each OFDM symbol is transmitted for a fixed time TFFT with
no tapering at the ends of the symbol. This symbol time corresponds
to the inverse of the subcarrier spacing of 1/TFFT Hz. The rectangu-
lar, boxcar waveform in the time domain results in a sinc frequency
response in the frequency domain. The sinc shape has a narrow main
lobe with many side lobes that decay slowly with the magnitude of
the frequency difference away from the center. Each carrier has a peak
at the center frequency and nulls evenly spaced with a frequency gap
equal to the carrier spacing [113].
The orthogonal nature of the transmission is a result of the peak
of each subcarrier corresponding to the nulls of all other subcarriers.
When this signal is detected using an FFT (fast Fourier transform),
the spectrum is not continuous but has discrete samples. If the FFT
is time synchronized, the frequency samples of the FFT correspond
to just the peaks of the subcarriers; thus, the overlapping frequency
region between subcarriers does not affect the receiver. The measured
peaks correspond to the nulls for all other subcarriers, resulting in
orthogonality between the subcarriers [113–115].
94 Im ag e En c ry p ti o n

7.3  OFDM System Model

This section briefly reviews the key steps in an OFDM communica-

tion system that can be used for the transmission of encrypted images.
The first step is the image encryption; the general model of a typi-
cal encryption system could be described with the following equation
[131]:
E(P, K) = C (7.3)

where P is the plainimage, E is the encryption algorithm, K is the

encryption key, and C is the cipherimage. The cipherimage is trans-
mitted through the communication channel. At the receiver side, the
decryption procedure could be represented by [131]

D(C, K’) = P’ (7.4)

where D is the decryption algorithm, K’ is the decryption key (it may
or may not be the same as the encryption key K), and P’ is the recov-
ered plainimage.
At the transmitter side, the encrypted data are converted into par-
allel data of N subchannels. Then, the data of each parallel subchan-
nel are modulated using a modulation scheme like PSK or QPSK
(quaternary PSK). For QPSK of N subchannels (d 0, d1, d2, … , dN−1),
each dn is a complex number d n = d I n + jdQn, where d I n and dQn are
{1, −1} [131–135].

7.3.1 FFT-OFDM

In FFT-OFDM, the modulated data are fed into an inverse FFT

(IFFT) circuit, and the OFDM signal is generated. The OFDM
splits a high-data-rate sequence into a number of low-rate sequences
that are transmitted over a number of subcarriers equal to N. The N
subcarriers are chosen to be orthogonal; that is, fn = n∆f, where ∆f =​
1/Ts, and Ts is the OFDM symbol duration. The resulting signal can
be expressed as follows [136]:

N −1 j 2 πkn

∑
1
x (n ) = X (k)e N , 0 ≤ t ≤ Ts (7.5)
N k =0

where X(k) represents the discrete-time samples.

 O r t h o g o n a l F req uen cy Di v isi o n Multip l e x in g 95

A guard interval is added at the start of each OFDM symbol to

eliminate the ISI, which occurs in multipath channels. An OFDM
symbol is extended in a cyclic manner to avoid the ICI. As a result, a
channel that is highly frequency selective is transformed into a large
set of individual flat fading, non-frequency-selective, narrowband
channels. At the receiver, the guard interval is removed, and the time
interval [0, Ts] is evaluated.

7.3.2 DCT-OFDM

The structure of the discrete cosine transform (DCT)-OFDM system

is similar to that of FFT-OFDM but with the IFFT and the FFT
modules replaced by inverse discrete cosine transform (IDCT) and
DCT modules, respectively. The main advantage of the DCT lies in
its excellent spectral energy compaction property, which makes most
of the samples transmitted close to zero, leading to a reduction in
the ISI. In addition, it uses only real arithmetic rather than the com-
plex arithmetic used in the FFT. This reduces the signal processing
complexity and the in-phase/quadrature imbalance [136,137]. In the
DCT-OFDM system, the transmitted signal is given by [138–140]

N −1
 πk ( 2n + 1) 
x (n ) =
2
N ∑ X ( k )β ( k ) cos 
k=0
2N  , p = 0,..., N − 1 (7.6)

where X(k) is the kth symbol of the input signal. β(k) can be written
as follows:

 1
 k=0 (7.7)
β( k ) =  2
 1 k = 1, 2,......., N − 1


7.3.3  Discrete Wavelet Transform–OFDM

The OFDM requires a cyclic prefix to remove ISI. This causes over-
head, and this overhead may sometimes be too large for the system
to be effective. The use of the wavelet transform reduces the ISI and
ICI [138, 141,142]. The structure of the discrete wavelet transform
96 Im ag e En c ry p ti o n

(DWT)-OFDM s­ ystem is similar to that of FFT-OFDM but with

the IFFT and the FFT modules replaced by indirect discrete wavelet
transform (IDWT) and DWT modules, respectively. The transmit-
ted signal with DWT-OFDM is given by [138]

x(n ) = ∑ X (k)ϕ(t − kT )
k=0
(7.8)

where φ(t) is the wavelet basis function.

Further advantages of the DWT-OFDM are as follows:
1. It requires less overhead as it does not require a cyclic perfix
(CP).
2. It does not require a pilot tone, which takes about 8% of the
subbands.
3. DWT-OFDM is inherently robust to ISI and ICI.
The DWT is explained in the following section.

7.3.4  Discrete Wavelet Transform

Wavelets have become a popular tool in most signal-processing

and communications applications. The conventional DWT may be
regarded as equivalent to filtering the input signal with a bank of
band-pass filters whose impulse responses are all approximately given
by scaled versions of a mother wavelet. The scaling factor between
adjacent filters is usually 2:1, leading to octave bandwidths and center
frequencies that are one octave apart [138,139]. The outputs of the
filters are usually maximally decimated so that the number of DWT
output samples equals the number of input samples, and the trans-
form is invertible as shown in Figure 7.2.

7.3.4.1 Implementation of the DWT  The DWT is normally imple-

mented by digital filters as shown for the one-dimensional (1D) case
in Figure  7.2. The art of finding a good wavelet lies in the design
of the set of filters H0, H1, G 0, and G1 to achieve various trade-offs
between spatial and frequency domain characteristics while satisfy-
ing the perfect reconstruction (PR) condition [142]. In Figure  7.2,
 O r t h o g o n a l F req uen cy Di v isi o n Multip l e x in g 97

X0(z) (1/2){X0(z)+ X0(–z)}

H0(z) 2 2 G0(z)

X(z) + Y(z)

H1(z) 2 2 G1(z)

X1(z) (1/2){X1(z)+ X1(–z)}

Figure 7.2  The two-band decomposition-reconstruction wavelet filter bank.

the process of decimation and interpolation by 2:1 at the output of H0

and H1 effectively sets all odd samples of these signals to zero.
For the low-pass branch, this is equivalent to multiplying x 0(n) by
1
2
( )
1 + ( −1) . Hence, X0(z) is converted to {X0(z) + X0(−z)}. Similarly,
n

1
X1(z) is converted to { X 1 ( z ) + X 1 ( − z )} .
2
Thus, the expression for Y(z) is given by [142]

1 1
Y (z) = { X 0 ( z ) + X 0 ( − z )}G 0 ( z ) + { X 1 ( z ) + X 1 ( − z )}G1 ( z )
2 2
1
= X ( z ) { H 0 ( z )G 0 ( z ) + H 1 ( z )G1 ( z )}
2
1
+ X ( − z ) { H 0 ( − z )G 0 ( z ) + H 1 ( − z )G1 ( z )} (7.9)
2

The first PR condition requires aliasing cancellation and forces this

term in X(−z) to be zero. Hence, {H0(−z)G 0(z) + H1(−z)G1(z)} = 0,
which can be achieved if [142]

H1(z) = z−aG 0(−z) and G1(z) = zaH0(−z) (7.10)

where a must be odd (usually a = ±1).

The second PR condition is that the transfer function from X(z) to
Y(z) should be unity [142]:

{H0(z)G 0(z) + H1(z)G1(z)} = 2 (7.11)

98 Im ag e En c ry p ti o n

If we define a product filter P(z) = H0(z)G 0(z) and substitute from

Equation (7.10) into Equation (7.11), then the PR condition becomes
[142]

H0(z)G 0(z) + H1(z)G1(z) = P(z) + P(−z) = 2 (7.12)

This needs to be true for all z, and because the odd powers of z in
P(z)cancel with those in P(−z), it requires that p 0 = 1 and pk = 0 for
all n even and nonzero. The polynomial P(z) should be a zero-phase
polynomial to minimize distortion. In general, P(z) is of the follow-
ing form [142]:

P(z) = … + p5z5 + p3z3 + p1z + 1 + p1z−1 + p3z−3 + p5z−5 + … (7.13)

The design method for the PR filters can be summarized in the

following steps [142]:
1. Choose p1, p3, p5, … to give a zero-phase polynomial P(z)
with good characteristics.
2. Factorize P(z) into H0(z) and G 0(z) with similar low-pass
­frequency responses.
3. Calculate H1(z) and G1(z) from H0(z) and G 0(z).
To simplify this procedure, we can use the following relation:

P(z) = Pt(Z) = 1 + Pt,1Z + Pt,3Z3 + Pt,5Z5 + … (7.14)

where

1
Z=
2
( )
z + z −1 (7.15)

7.3.4.2  Haar Wavelet Transform  The Haar wavelet is the simplest

type of wavelet. In discrete form, Haar wavelets are related to a
mathematical operation called the Haar transform. The Haar trans-
form serves as a prototype for all other wavelet transforms. Like
all wavelet transforms, the Haar transform decomposes a discrete
signal into two subsignals of half its length. One subsignal is a run-
ning average or trend; the other subsignal is a running difference or
fluctuation.
 O r t h o g o n a l F req uen cy Di v isi o n Multip l e x in g 99

The Haar wavelet uses the simplest possible Pt(Z) with a single zero
at Z = −1. It is represented as follows [142]:

1
Pt(Z) = 1 + Z and Z =
2
( )
z + z −1 (7.16)

Thus,

1 1
P(z ) =
2
( )
z + 2 + z −1 = ( z + 1) 1 + z −1
2
( )
= G0 ( z ) H 0 (z) (7.17)

We can find H0(z) and G 0(z) as follows:

1
H 0 (z) =
2
( )
1 + z −1 (7.18)

G 0(z) = (z + 1) (7.19)
Using Equation (7.10) with a = 1, we obtain

1 1
G1 ( z ) = zH 0 ( − z ) =
2
( )
z 1 − z −1 = ( z − 1) (7.20)
2

H1(z) = z−1G 0(−z) = z−1(−z + 1) = (z−1 − 1) (7.21)

7.4  Guard Interval Insertion

The guard interval insertion is an important step in all OFDM sys-

tems. The purpose of this step is to avoid the ISI. Figure 7.3 illustrates
the process of guard interval insertion [143–146]. Also, the guard
period may consist of two sections: zero-amplitude transmission
data and a cyclic extension of the transmitted symbol. We consider
all these types of guard interval and their effects on the transmitted
encrypted images.

OFDM Symbol Guard OFDM Symbol Guard OFDM Symbol

Figure 7.3  Guard interval insertion in OFDM.

10 0 Im ag e En c ry p ti o n

7.5  Communication Channels

7.5.1  Additive White Gaussian Noise Channel

For an additive white Gaussian noise (AWGN) channel, the received

signal r(t) is expressed as [141]
r(t) = s(t) + n(t) (7.22)
where s(t) is the transmitted signal, r(t) is the received signal, and n(t)
is the AWGN with power spectral density given by [141]
Φ(f ) = N0/2 [W/Hz] (7.23)
where N0 is a constant that is often called the noise power density.

7.5.2  Fading Channel

The path between the transmitter and receiver is characterized by vari-

ous obstacles and reflections, which lead to a fading effect at the receiver
as shown in Figure 7.4. We can take the following expression to simulate
the incoming signal at the receiver in the Raleigh fading scenario [141]:
f(t) = x(t) + j · y(t) (7.24)

 N1
  2πn  
=
2
 N 1 + 1
∑ sin  Nπn  cos 2πf
n =1
1
d cos  t
 N 1  

1 
+ cos(2πf d t ) 
N1 + 1 

Reflected
Paths

Direct Paths Transmitter

Figure 7.4  A wireless channel with multipath propagation.

 O r t h o g o n a l F req uen cy Di v isi o n Multip l e x in g101

N1
  2πn  
+j
2
N1 ∑ sin  Nπn  cos 2πf
n =1
1
d cos  t
 N 1  
(7.25)

where the wave number of the incoming wave is N0, and N1 is given by

N1 = 1/2((N0/2) – 1) (7.26)

7.6  Channel Estimation and Equalization

Channel estimation and equalization constitute a major challenge in

current and future communication systems. The equalizers utilized
to  compensate for the ISI can be classified as linear and nonlinear
equalizers [147]. Linear equalizers are simple linear filter structures that
try to invert the channel in the sense that the product of the transfer
functions of the channel and the equalizer fulfills a certain criterion.
This criterion can either be achieving a completely flat transfer function
of the channel filter concatenation or minimizing the mean square error
at the filter output [148]. One of the most popular linear equalizers is
the zero-forcing (ZF) equalizer. The ZF solution can be written as [148]

WZF = (HHH)−1HH (7.27)

where H is the matrix of the channel transfer function.

The advantage of the ZF equalizer is that the statistics of the addi-
tive noise and source data are not required.
Channel estimation can be performed with pilot symbol estima-
tion (PSE). With an OFDM system, the wideband channel is divided
into a number of narrowband channels. Thus, channel estimation can
be performed by inserting pilot symbols with a known modulation
scheme into the transmitted signal. Based on these pilot symbols, the
receiver can measure the channel transfer function for each subcarrier
using interpolation techniques [113].
The mathematical model for the PSE and equalization can be
­represented as follows [144]:

 i1   i0 
  = A  (7.28)
 q1   q0 
10 2 Im ag e En c ry p ti o n

where i1 is the received in-phase symbol, q1 is the quadrature received

symbol, i0 is the transmitted in-phase symbol, q 0 is the quadrature
transmitted symbol, and A is the transition matrix of the fading envi-
ronment that is given by [130]

A= ( a11
a21
a12
a22 ) (7.29)

The channel matrix is Toplitz with a11 = a22 and a12 = −a21. To per-
form channel equalization, we need to estimate the matrix A−1 as
follows [130]:

A −1 = 2
a +a
11
1
2
12
( a11
− a12
a12
a11 ) (7.30)

Because the transmitted and received symbols are known, we can

estimate the channel coefficients using Equations (7.28) and (7.30) as
follows [130]:

1
a11 = a 22 =
i1 + q1
(i0i1 + q0q1 ) (7.31)

1
a 21 = − a12 =
i1 + q1
(q0i1 + i0q1 ) (7.32)

Once the channel is estimated, it can be used for equalization

with unknown transmitted symbols [113]. Equalization is not the
only problem encountered in wireless communication of images with
OFDM. There are some other limitations that may affect the quality
of transmitted images, especially if they are encrypted. Some of these
problems are addressed in the next chapter.
 8
OFDM L imitati ons

8.1 Introduction

OFDM has been adopted in the European digital audio and video
broadcasting radio system and is being investigated for broad-
band indoor wireless communications. Standards such as High-
Performance Radio Local-Area Network (HIPERLAN2), Institute
of Electrical and Electronics Engineers (IEEE) 802.11a, and IEEE
802.11g have emerged to support s­ervices based on the Internet
Protocol (IP). Such systems are based on OFDM and are designed to
operate in the 5-GHz band [147].
Unfortunately, OFDM communication systems have two
­primary problems: the high sensitivity to carrier frequency o­ ffsets
(CFOs) and the high peak-to-average power ratio (PAPR). The
sensitivity to CFOs breaks the subcarriers’ orthogonality, and the
high PAPR requires system components with a wide l­inear range
to ­accommodate for the signal variations. Otherwise, ­nonlinear
d istortion, which results in a loss of subcarrier orthogonal-
­
ity and  hence a degradation in the system performance, occurs
[148–152].
Researchers have proposed various methods to combat the
intercarrier interference (ICI in OFDM systems. The existing
­
approaches that have been developed to reduce ICI can be catego-
rized into frequency domain equalization, time domain windowing,
and self-cancellation (SC) schemes. In addition, statistical approaches
have been explored to estimate and cancel ICI. In this chapter, the
effects of CFO and PAPR problems are studied with solutions to these
­problems and to reduce their effects on the transmitted encrypted
images.

10 3
10 4 Im ag e En c ry p ti o n

8.2  Analysis of Intercarrier Interference

The main problem of OFDM is its susceptibility to small differences

in frequency at the transmitter and receiver, normally referred to as
frequency offset. This frequency offset can be caused by Doppler shift
due to relative motion between the transmitter and receiver or by dif-
ferences between the frequencies of the local oscillators at the trans-
mitter and receiver. The frequency offset is modeled as a multiplicative
factor introduced in the channel, as shown in Figure 8.1) [153–155].
The received signal is given by

 j 2πnε 
y (n ) = i (n ) exp  + w (n ) (8.1)
 N 

where ε is the normalized frequency offset and is given by ΔfNTs. Δf is

the frequency difference between the transmitted and received carrier
frequencies, and Ts is the subcarrier symbol period. w(n) is the addi-
tive white Gaussian noise (AWGN) introduced in the channel.
The effect of this frequency offset on the received symbol stream
can be understood by considering the received symbol Y(k) on the kth
subcarrier [151–153].

N −1

Y ( k ) = X ( k )S (0) + ∑ X (l )S(l − k) + W (k) (8.2)

l = 0,l ≠ k

where k = 0, 1, … , N − 1; N is the total number of subcarriers; X(k) is

the transmitted symbol for the kth subcarrier; W(k) is the fast Fourier
transform (FFT) of w(n); and S(l − k) are the complex coefficients for
the ICI components in the received signal. The ICI components are

j2πnε
exp( ) w(n)
N

i(n) y(n)

Figure 8.1  Frequency offset model.

 OFDM Limitati o ns 10 5

the interfering signals transmitted on subcarriers other than the kth

subcarrier. The complex coefficients are given by [151–153]

sin( π(l + ε − k )) 1
S (l − k ) = exp( jπ(1 − )(l + ε − k )) (8.3)
N sin( π(l + ε − k ) / N ) N

8.3  CFO in FFT-OFDM System Model

In FFT-OFDM, the modulated symbols are grouped into blocks,

each containing N symbols, and the inverse FFT (IFFT) is per-
formed. The resulting signal after the IFFT can be expressed as fol-
lows [138,151–153]:

N −1 j 2 πnk
x(n ) =
1
N ∑k=0
X ( k )e Ts
, 0 ≤ n ≤ N − 1 (8.4)

where N is the number of subcarriers. X(k) represents the kth modu-

lated symbol.
At the end of the transmitter, a guard interval is inserted between
symbols with hybrid guard interval or zero padding to eliminate the
intersymbol interference (ISI). The resulting signal is then transmit-
ted through the wireless channel.
At the receiver side, the padded zeros or the hybrid guard interval
are removed from the received signal. The received signal in the pres-
ence of CFO is given by

j 2 πmε
r (n ) = x(n )e N + w(n ) (8.5)

where ε = ΔfTs is the normalized CFO, Δf is the CFO, Ts is the

OFDM symbol duration, and w(n) is AWGN. After that, the
received signal is transformed into the frequency domain via the
FFT as follows:

N −1  −2 πnk 
R( k ) = ∑ r (n)e
n=0
 N 
(8.6)
10 6 Im ag e En c ry p ti o n

Substituting Equation (8.3) into Equation (8.4), we obtain

[151–153]

N −1  2 πmε   −2 πnk 
R( k ) = ∑ x(n)e
n=0
 N   N 
e + W (k)

N −1 N −1  2 πn ( l +ε − k )  (8.7)
∑ X (l )∑ e
1  
= N
+ W (k)
N l =0 n=0

where W(k) is the kth frequency domain sample of the noise. We

N −1  j 2 πn ( l −ε − k ) 

∑e
1  
can expand N
using the geometric series as follows
N n=0
[151–153]:

N −1  j 2 πn ( l −ε− k ) 
1 1 − e ( j 2 π ( l +ε− k ))
1
∑
 
N
e =  j 2 π ( l +ε− k ) 
N N  
n=0
1 − e N 
(8.8)
1 sin( π(l − k + ε ))
= e ( jπ ( l +ε− k )(1−1/ N ))
N sin( π(l − k − ε ) / N )

We obtain [151–153]

N −1

R( k ) = ∑ X (l )S(l − k + ε ) + W (k) (8.9)

l =0

where S(l − k + ε) represents the complex coefficients for the inter-

ference components in the received signal. S(l − k + ε) is given by
[151–153]

sin( π(l − k + ε ))
S (l − k + ε ) = e ( jπ ( l +ε− k )(1−1/ N )) (8.10)
M sin( π(l − k − ε ) / N )
 OFDM Limitati o ns 10 7

8.4  CFO in DCT-OFDM System Model

The main advantage of the discrete cosine transform (DCT) lies in

its excellent spectral energy compaction property, which makes most
of the samples transmitted close to zero, leading to a reduction in the
ISI. In addition, it uses only real arithmetic rather than the complex
arithmetic used in the FFT. This reduces the signal-processing com-
plexity, and the in-phase/quadrature imbalance [150,151]. In the DCT-
OFDM system, the transmitted signal is given by [136,137,139–143]

N −1

x(n ) =
2
N ∑ X (k)β(k)cos  πk(22nN+ 1)  ,
k=0
p = 0,..., N − 1 (8.11)

where X(k) is the kth symbol of the input signal. β(k) can be written
as follows:

 1
 k=0
β( k ) =  2 (8.12)
 1 k = 1, 2,......., N − 1


In the DCT-OFDM system, the complex coefficients for the inter-

ference components in the received signal can be expressed as follows
[151–153]:

1
S (l , k , ζ ) = β kβl [ φ(l + k − ζ) + φ(l − k − ζ)
2N
+φ(l + k + ζ) + φ(l − k + ζ)] (8.13)

+ j [ Γ(l + k − ζ ) + Γ(l − k − ζ )

−Γ(l + k + ζ) − Γ(l − k + ζ)]

where ζ = 2T∆f is the normalized CFO for the DCT-OFDM system

[151–153].
10 8 Im ag e En c ry p ti o n

2
 πx 
sin  
 2
Γ( x ) = (8.14)
 πx 
sin 
 2 N 

 πx   πx 
sin   cos  
 2  2
φ( x ) = (8.15)
 π x 
sin 
 2 N 

8.5  CFO in DWT-OFDM System Model

The transmitted signal with discrete wavelet transform (DWT)-

OFDM is given by [138]
N

x(n ) = ∑ X (k)ϕ(t − kT )
k=0
(8.16)

where φ(t) is the wavelet basis functors. Some attempts have been
made to predict the CFO effect on DWT-OFDM, but, as of yet, no
closed-form expression has been presented for this task.

8.6  CFO Compensation

There are many techniques that were developed to compensate for the
CFO in multicarrier communication systems. In this chapter, the com-
pensation process is carried out in the time domain. Mathematically,
for the OFDM system, the received sequence is multiplied by a time
j 2 πnε
−
domain sequence e N before the FFT/DCT/DWT processing. The
compensated signal can be formulated as follows for AWGN [151–153]:

j 2 πnε j 2 πnε
− −
rc (n ) = e N r (n ) = x (n ) + e N w(n ) (8.17)

In fading channels, as we know, estimation of the channel trans-

fer function is needed. The channel transfer function estimate may
be computed using any algorithm that gives reliable estimates. In our
design, the zero-forcing (ZF) channel estimator was applied to obtain
 OFDM Limitati o ns 10 9

the initial channel estimate when the first OFDM symbol of the
superframe is received. This symbol represents the pilot symbol known
to the receiver. After receiving the first OFDM symbol, the estimator
switches to the tracking mode. The channel estimates are refined and
tracked according to the gradient algorithm.

8.7  Simulation Parameters

In our simulations, we carried out the encryption of the Cameraman

image with the previously discussed encryption algorithms. Then, the
encrypted images were transformed into a binary format and used for
OFDM modulation. Both AWGN and Raleigh fading channels were
considered in the simulations. A normalized CFO was used in some
of the simulation experiments. PAPR reduction techniques were used
in some other experiments. In all experiments, 128 subcarriers and
quaternary phase shift keying (QPSK) were used. The simulation
parameters are summarized in Table 8.1.
To evaluate the quality of the decrypted images at the receiver,
we used the peak signal-to-noise ratio (PSNR) between the original
image and the decrypted image, which is defined as follows:

 255 2 
PSNR = 10 log10  (8.18)
 MSE 

where the mean square error (MSE) is defined as

U U
1
∑ ∑  f (i, j ) − fˆ (i, j )
2
MSE = (8.19)
U2 i =1 j =1

Table 8.1  Simulation Parameters

PARAMETER DESCRIPTION
Transmitter System bandwidth 64 MHz
Modulation type QPSK
cyclic prefix length 16 samples
N 128 subcarriers
Subcarrier spacing 500 kHz
Type of transform FFT, DCT, or DWT
Channel Channel model Raleigh fading
Receiver PSE and equalizer ZF equalization
Encryption algorithm Type of encryption Chaotic encryption, chaotic-OFB encryption
110 Im ag e En c ry p ti o n

where f(i, j) is the original image of dimensions U × U and fˆ (i , j ) is

the decrypted image.
Figures 8.2 and 8.3 illustrate the PSNR performance versus the Eb /
N0 for the FFT-OFDM, the DCT-OFDM, and the DWT-OFDM
systems with chaotic and chaotic-output feedback (OFB) encryption
algorithms in a frequency-selective channel, respectively. Without
CFO compensation and with pilot symbol estimation (PSE), it was
shown that CFO disrupted the PSNR performance of all OFDM
systems; with CFO compensation, the PSNR performances of all
systems were better than those without CFO compensation. These
figures also show that the PSNR performances of all encryption algo-
rithms with the FFT-OFDM s­ystem were the best compared with
those of the DCT-OFDM and the DWT-OFDM systems.
Figure  8.4 gives the output of the FFT-OFDM system for the
transmission of the original Cameraman image over an AWGN
­channel at N = 128 and G = 32.
Figure  8.5 gives the output of the FFT-OFDM system for the
transmission of the original Cameraman image over a fading channel
at N = 128 and G = 32.

80

FFT with CFO

70 DCT with CFO
DWT with CFO
60 FFT with PSE
DCT with PSE
50 DWT with PSE
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 8.2  PSNR versus Eb /N0 for chaotic-encrypted image transmission with OFDM system over
a Raleigh channel.
 OFDM Limitati o ns 111

80
FFT with CFO
70 DCT with CFO
DWT with CFO
60 FFT with PSE
DCT with PSE
DWT with PSE
50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 8.3  PSNR versus Eb /N0 for chaotic-OFB-encrypted image transmission with OFDM sys-
tem over a Raleigh fading channel.

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 14.6993 PSNR = 23.7838

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 48.0593 PSNR > 60

Figure 8.4  Original image with FFT-OFDM over an AWGN channel.

112 Im ag e En c ry p ti o n

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 17.8214 PSNR = 29.4547

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 53.8719 PSNR > 60

Figure 8.5  Original image with FFT-OFDM over a fading channel.

Figure  8.6 gives the output of the FFT-OFDM system for the
transmission of the chaotic-encrypted image over an AWGN channel
at N = 128 and G = 32.
Figure  8.7 gives the output of the FFT-OFDM system for the
transmission of the chaotic-encrypted image over a fading channel at
N = 128 and G = 32.
Figure  8.8 gives the output of the DCT-OFDM system for the
transmission of the original Cameraman image over an AWGN
­channel at N = 128 and G = 32.
Figure  8.9 gives the output of the DCT-OFDM system for the
transmission of the original Cameraman image over a fading channel
at N = 128 and G = 32.
 OFDM Limitati o ns 113

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 15.1527 PSNR = 25.1709

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 50.5529 PSNR >60

Figure 8.6  Chaotic-encrypted image with FFT-OFDM over an AWGN channel.

Figure 8.10 gives the output of the DCT-OFDM system for the

transmission of the chaotic-encrypted image over an AWGN channel
at N = 128 and G = 32.
Figure 8.11 gives the output of the DCT-OFDM system for the
transmission of the chaotic-encrypted image over a fading channel at
N = 128 and G = 32.
Figure  8.12 gives the output of the DWT-OFDM system for
the transmission of the original Cameraman image over an AWGN
channel at N = 128 and G = 32.
Figure 8.13 gives the output of the DWT-OFDM system for the
transmission of the original Cameraman image over a fading channel
at N = 128 and G = 32.
114 Im ag e En c ry p ti o n

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 18.3597 PSNR = 30.0828

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 53.8881 PSNR >60

Figure 8.7  Chaotic-encrypted image with FFT-OFDM over a fading channel.

Figure 8.14 gives the output of the DWT-OFDM system for the

transmission of the chaotic-encrypted image over an AWGN channel
at N = 128 and G = 32.
Figure 8.15 gives the output of the DWT-OFDM system for the
transmission of the chaotic-encrypted image over a fading channel at
N = 128 and G = 32.
Table 8.2 summarizes these results.

8.8  Effect of PAPR

One of the main limitations of OFDM is the high PAPR. A ­signal

with large peaks can be obtained by the constructive ­superposition
of subcarriers. The PAPR is linearly dependent on the number of
 OFDM Limitati o ns 115

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 15.7172 PSNR = 26.3188

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 53.5993 PSNR >60

Figure 8.8  Original image with DCT-OFDM over an AWGN channel.

s­ubcarriers. High-power peaks make certain demands on power

amplifiers and analog-to-digital (A/D) and digital-to-analog (D/A)
converters. Large peaks are distorted nonlinearly due to the power
amplifier imperfections, and intermodulation products appear. They
can be interpreted as ICI and out-of-band radiation.
This high PAPR results in significant in-band distortion and
out-of-band radiation when the signal passes through a nonlinear
device. The in-band distortion increases the bit error rate (BER), and
the ­out-of-band radiation results in unacceptable adjacent channel
interference (ACI) [152]. Without the use of any PAPR reduction
technique, the efficiency of power consumption at the transmitter
becomes very poor. Because OFDM signals are modulated indepen-
dently in each ­subcarrier, the combined OFDM signals are likely to
have large peak powers at certain instances. The peak power increases
as the n
­ umber of s­ubcarriers increases. The peak power is generally
116 Im ag e En c ry p ti o n

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 16.7843 PSNR = 23.2367

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 27.9452 PSNR = 29.3708

Figure 8.9  Original image with DCT-OFDM over a fading channel.

evaluated in terms of the PAPR. The PAPR of the transmitted

OFDM signal is defined as

PAPR =
Ppeak
=
max xm ( ) 2

, m = 0,1,......., N − 1 (8.20)
N
Pavg
1/ N xm
2
∑
m=0

In Equation (8.20), the numerator represents the maximum envelope

power, and the denominator represents the average power.
The cumulative distribution function (CDF) of the PAPR is one of
the most frequently used performance metrics for PAPR performance
before and after applying PAPR reduction techniques. In the litera-
ture, the complementary CDF (CCDF) is commonly used instead of
 OFDM Limitati o ns 117

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 15.0481 PSNR = 24.4613

(c) Eb/N0 = 10 dB (d) Eb/N0= 15 dB

PSNR = 53.4971 PSNR >60

Figure 8.10  Chaotic-encrypted image with DCT-OFDM over an AWGN channel.

the CDF itself. The CCDF of the PAPR denotes the probability that
the PAPR of a data block exceeds a given threshold. In [153], a simple
approximate expression was derived for the CCDF of the PAPR of
a multicarrier signal; the real and imaginary values of x(t) follow a
Gaussian distribution. The CDF of the amplitude z of an OFDM
signal sample is given by
F(z) = 1 − e−z (8.21)
We need to derive the CDF of the PAPR for an OFDM data
block. Assuming that the signal samples are mutually independent
and uncorrelated, the CDF of the PAPR for an OFDM data block
can be found as
P(PAPR ≤ z) = F(z)N = (1 − e−z)N (8.22)
118 Im ag e En c ry p ti o n

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 16.3858 PSNR = 23.0375

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 30.3995 PSNR = 33.2591

Figure 8.11  Chaotic-encrypted image with DCT-OFDM over a fading channel.

The assumption made that the signal samples are mutually inde-
pendent and uncorrelated is not true when oversampling is applied.
Also, this expression is not accurate for a small number of subcar-
riers because the Gaussian assumption does not hold in this case
[156–159]. It was shown that the PAPR of an oversampled signal for
N subcarriers is approximated by the distribution for αN subcarriers
without oversampling, where α is larger than 1 [160]. In other words,
the effect of oversampling is approximated by adding a certain num-
ber of extra signal samples.
The CDF of the PAPR of an oversampled signal is thus given by

P(PAPR ≤ z) ≈ (1 − e−z)αN (8.23)

 OFDM Limitati o ns 119

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 14.6839 PSNR = 23.7368

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 45.9748 PSNR >60

Figure 8.12  Original image with DWT-OFDM over an AWGN channel.

It was found that α = 2.3 is a good approximation for four-times

oversampled OFDM signals [161]. According to Equation (8.23), the
probability that the PAPR exceeds PAPR0 is given by

CCDF = Pr[ PAPR > PAPR0 ] = 1 − (1 − e PAPR0 )αN (8.24)

8.9  PAPR Reduction Methods

Several methods have been presented in the literature for PAPR

reduction of OFDM signals. In this book, we concentrate on three
such methods: clipping, companding, and hybrid clipping and
­companding. These are three simple PAPR reduction methods. Our
objective of is to show the effect of PAPR reduction methods on
encrypted image transmission.
12 0 Im ag e En c ry p ti o n

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 17.5731 PSNR = 24.9482

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 32.6435 PSNR = 37.7109

Figure 8.13  Original image with DWT-OFDM over a fading channel.

8.9.1  The Clipping Method

Clipping is the simplest method that can limit the amplitude of the
signal to some desired threshold value. It is a nonlinear process and
may cause in-band distortion and out-of-band radiation. This leads
to degradation in the BER performance [162,163]. In the clipping
method, when the amplitude exceeds a certain threshold, it is hard
clipped, while the phase is saved. Namely, when we assume the phase
of the baseband OFDM signal xm is ϕm and the threshold is A, the
output signal after clipping will be given as follows:

 xm > A  (8.25)
 Ae jφm
xm =  
 xm xm ≤ A 

 OFDM Limitati o ns 121

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 14.7090 PSNR = 23.5544

(c) Eb/N0= 10 dB (d) Eb/N0 = 15 dB

PSNR = 44.6552 PSNR >60

Figure 8.14  Chaotic-encrypted image with DWT-OFDM over an AWGN channel.

The clipping ratio (CR) is a useful means to represent the clipping

level. It is the ratio between the maximum power of the clipped signal
and the average power of the unclipped signal. If the IFFT/inverse
direct cosine transform (IDCT)/inverse direct wavelet transform
(IDWT) output signal is normalized, the unclipped signal power is
1. If we clip the IFFT/IDCT/IDWT output at level A, then the CR
is (A)2 / 1 = (A)2.

8.9.2  The Companding Method

The companding transform uses a compander to reduce the signal

amplitude. Such an approach can effectively reduce the PAPR with
less computational complexity [164]. The corresponding ­transmitter
and receiver need a compander and an expander, respectively, which
12 2 Im ag e En c ry p ti o n

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 16.3689 PSNR = 23.8108

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 32.7138 PSNR = 37.5838

Figure 8.15  Chaotic-encrypted image with DWT-OFDM over a fading channel.

of course slightly increases the hardware cost. In simulation experi-

ments, compression is performed according to the well-known µ-law.
The compression process can be expressed as follows [164,165]:
ln[1 + µ xm / V max ]
xc = V max sgn[ xm ] (8.26)
ln[1 + µ ]

where
V max = max( xm ) m = 0,1,.....N − 1

where μ is the companding coefficient, xc is the companded sample,
and xm is the original sample. The expansion process is simply the
inverse of Equation (8.26):
∧ V max x ln(1 + µ )
xm = [exp( c − 1)]sgn( xc )
µ V max
 OFDM Limitati o ns 12 3

Table 8.2  Summarized Results

Eb/N0 Eb/N0 = 0 dB Eb/N0 = 5 dB Eb/N0 = 10 dB

ORIGINAL + AWGN CHANNEL

PSNR with FFT 14.6993 23.7838 48.0593
PSNR with DCT 15.7172 26.3188 53.5993
PSNR with DWT 14.6839 23.7368 45.9748
ORIGINAL + FADING CHANNEL
PSNR with FFT 17.8214 29.4547 53.8719
PSNR with DCT 16.7843 23.2367 27.9452
PSNR with DWT 17.5731 24.9482 32.6435
CHAOTIC + AWGN CHANNEL
PSNR with FFT 15.1527 25.1709 50.5529
PSNR with DCT 15.0481 24.4613 53.4971
PSNR with DWT 14.7090 23.5544 44.6552
CHAOTIC + FADING CHANNEL
PSNR with FFT 18.3597 30.0828 53.8881
PSNR with DCT 16.3858 23.0375 30.3995
PSNR with DWT 16.3689 23.8108 32.7138
DES + AWGN CHANNEL
PSNR with FFT 8.3766 11.1743 36.4777
PSNR with DCT 8.3762 11.1133 34.4818
PSNR with DWT 8.3954 11.2270 32.2373
DES + FADING CHANNEL
PSNR with FFT 9.6637 16.7520 41.4821
PSNR with DCT 9.0390 11.7067 16.5200
PSNR with DWT 8.4729 10.7940 17.4856
AES + AWGN CHANNEL
PSNR with FFT 8.4026 9.4551 31.5464
PSNR with DCT 8.4211 9.4006 30.5205
PSNR with DWT 8.4049 9.3407 28.8218
AES + FADING CHANNEL
PSNR with FFT 8.6932 14.1673 36.8018
PSNR with DCT 8.5829 10.2435 14.1012
PSNR with DWT 8.3737 9.2508 13.7209
RC6 + AWGN CHANNEL
PSNR with FFT 8.3916 9.4165 30.9051
PSNR with DCT 8.3961 9.3528 33.1261
PSNR with DWT 8.4344 9.3419 29.1988
RC6 + FADING CHANNEL
PSNR with FFT 8.7540 13.8893 40.0734
PSNR with DCT 8.5691 10.2276 14.1490
PSNR with DWT 8.3983 9.2497 13.9828
AES, Advanced Encryption Standard; DES, Data Encryption Standard.
12 4 Im ag e En c ry p ti o n

∧
where x m is the estimated sample after expansion.

8.9.3  The Hybrid Clipping-and-Companding Method

The hybrid clipping-and-companding method comprises clipping fol-

lowed by µ-law companding. By exploiting the clipping in the first
step, the hybrid scheme can reduce the PAPR of OFDM signals with
slight BER degradation. Moreover, the companding in the second
step further reduces the PAPR of the OFDM signals. Consequently,
we expect that the hybrid scheme effectively reduces the PAPR with
slight BER degradation, while the complexity of the system is slightly
increased.
A complete image transmission system with OFDM comprises a
PAPR reduction stage and an error-correction coding stage to reduce
the effect of PAPR reduction at the receiver. A convolutional code
with rate 1/2, constraint length 7, and octal generator polynomial
(133,171) is adopted in the simulation experiments.

8.10  Simulation Experiments of PAPR Reduction Methods

Several experiments were carried out to test the effect of the PAPR
reduction methods on the process of encrypted image transmission
with OFDM. The compared PAPR reduction methods are the clip-
ping, companding, and hybrid clipping-and-companding methods.
The performance of the process of chaotic image transmission with
encrypted (FFT/DCT/DWT)-OFDM systems having N = 128 and
G = 32 over an AWGN is shown in Figures 8.16 to 8.18. From these
figures, it is clear that the PSNR performance of the ­chaotic-encrypted
image transmission with FFT/DCT/DWT-OFDM is better with
µ = 0.1 than that with µ = 4 at Eb /N0 ≥ 10 dB. At high SNR values,
it is possible to receive the ­chaotic encrypted images with a very large
PSNR if error-­correction codes are utilized.
The performance of the process of encrypted image transmission
using the chaotic-OFB algorithm with (FFT/DCT/DWT)-OFDM
systems over an AWGN is shown in Figures 8.19 to 8.21. From these
figures, it is clear that the PSNR performance of the chaotic-OFB-
encrypted image transmission with FFT/DCT/DWT-OFDM is
better with µ = 0.1 than with µ = 4 at Eb /N0 ≥ 10 dB. At high SNR
 OFDM Limitati o ns 12 5

100
Without coding
90 With clipping
With companding at µ = 0.1
80 With clipping + companding at µ = 0.1
With companding at µ = 4
70 With clipping + companding at µ = 4
PSNR(dB)

60

50

40

30

20

10
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 8.16  PSNR versus Eb /N0 for chaotic-encrypted image transmission with FFT-OFDM sys-
tem over an AWGN channel. Convolutional coding is applied with all cases except the first one.

100

90 Without coding
With clipping
With companding at µ = 0.1
80 With clipping + companding at µ = 0.1
With companding at µ = 4
70 With clipping + companding at µ = 4
PSNR(dB)

60

50

40

30

20

10
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 8.17  PSNR versus Eb /N0 for chaotic-encrypted image transmission with DCT-OFDM sys-
tem over an AWGN channel. Convolutional coding is applied with all cases except the first one.
12 6 Im ag e En c ry p ti o n

100
Without coding
90 With clipping
With companding at µ = 0.1
80
With clipping + companding at µ = 0.1
70 With companding at µ = 4
With clipping + companding at µ = 4
60
PSNR(dB)

50

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 8.18  PSNR versus Eb /N0 for chaotic-encrypted image transmission with DWT-OFDM sys-
tem over an AWGN channel. Convolutional coding is applied with all cases except the first one.

100
Without coding
90 With clipping
With companding at µ = 0.1
80 With clipping + companding at µ = 0.1
With companding at µ = 4
70
With clipping + companding at µ = 4
60
PSNR(dB)

50

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 8.19  PSNR versus Eb /N0 for chaotic-OFB-encrypted image transmission with FFT-OFDM
system over an AWGN channel. Convolutional coding is applied with all cases except the first one.
 OFDM Limitati o ns 12 7

100
Without coding
90 With clipping
With companding at μ = 0.1
80
With clipping + companding at μ = 0.1
With companding at μ = 4
70
With clipping + companding at μ = 4
60
PSNR(dB)

50

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 8.20  PSNR versus Eb /N0 for chaotic-OFB-encrypted image transmission with DCT-OFDM
system over an AWGN channel. Convolutional coding is applied with all cases except the first one.

100
Without coding
90
With clipping
With companding at μ = 0.1
80
With clipping + companding at μ = 0.1
70 With companding at μ = 4
With clipping + companding at μ = 4
60
PSNR(dB)

50

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 8.21  PSNR versus Eb /N0 for chaotic-OFB-encrypted image transmission with DWT-OFDM
system over an AWGN channel. Convolutional coding is applied with all cases except the first one.
12 8 Im ag e En c ry p ti o n

values, it is ­possible to receive the chaotic-OFB-encrypted images

with a very large PSNR if error-correction codes are utilized. From
all obtained results, we can see that PAPR reduction technique effects
can be ­mitigated to a great extent with error-correcting codes such
as the convolutional code. Another conclusion is that chaotic-OFB
encryption preserves its rank as the best encryption algorithm from
the immunity to noise perspective even in the presence of PAPR
reduction. A simple clipping, companding, or hybrid clipping-and-
companding method may be feasible for PAPR reduction in the case
of encrypted image transmission.

8.11  Sampling Clock Offset

The detection of OFDM symbols cannot be done properly without

reliable clock synchronization. One synchronization step consists of
estimating the OFDM symbol timing, which is the delay between the
transmitted and the received OFDM symbols. In a certain number
of applications where these symbols are short, estimating this delay
is enough. However, as soon as the number of samples per OFDM
symbol (or, equivalently, the number of subcarriers) becomes large,
the frequency offset between the transmitter sampling clock and the
receiver sampling clock in its free oscillation mode also has to be con-
sidered. Indeed, this offset leads to a sampling delay that drifts lin-
early in time over the OFDM symbol. Without any compensation,
this drift hampers the receiver performance as soon as the product
of the relative clock frequency offset with the number of subcarriers
becomes nonnegligible in comparison with one [180]. For instance, in
very-high-speed digital subscriber line (VDSL) transmissions, these
two quantities can respectively reach 10 −4 and 4096 [167], making the
clock frequency offset compensation mandatory. As another example,
power line transmissions (PLTs) in the band (1 MHz, 20 MHz) [168]
show similar behavior with respect to this phenomenon.
As it is well known, the part of the OFDM symbol that enters the
FFT device at the receiver comes after a cyclic prefix. As the prefix
has a length comparable to the channel impulse response length, it
is precisely when the channel is long that a long duration has to be
chosen for the useful part of the OFDM symbol to reduce the impact
of the cyclic prefix on the spectral efficiency. It is therefore worth
 OFDM Limitati o ns 12 9

considering the problem of the joint estimation of the clock frequency

offset and of the channel impulse response, particularly when the
observation window has to be rather large [166–168].

8.11.1  System Model

Let us consider the reception of one standard OFDM block that has
passed through a nonflat fading channel. After removing the guard
interval, the observation window size is To = NT where N is the ­number
of subcarriers, T is the sampling period at the transmitter, and 1/To is
the spacing between two adjacent subcarriers. Consequently, the con-
tinuous-time received signal y N( a ) (t ) is written as follows:

y N( a ) (t ) = ∑d
k ∈Z
N .k g ( a ) (t − kT ) + v ( a ) (t ) (8.27)

where (dN,k)k = 0,1,....,N − 1 represents the output of the N-fold IFFT device

of the transmitter. This OFDM symbol is devoted to training and
therefore is assumed to be known at the receiver. As usual, N is a
power of 2. The unknown impulse response g(a)(t) represents the
­complete channel that includes the transmit filter, the propagation
channel, and the receiver low-pass filter. Finally. v(a)(t) is an additive
noise independent of the data. Because of the oscillators’ imperfec-
tion, the transmitter and receiver clocks are not synchronized.
Therefore, y N( a ) (t ) is sampled at (1 + δ)T instead of T, where δ is
an unknown offset lying in the known interval [−δmax, δmax]. The
parameter δmax is related to the precision of the oscillators used in the
transmission chain. The asymmetric digital subscriber line (ASDL)/
VDSL norms [167], for instance, r­ecommend that δmax be equal to
10 −4. The discrete time received ­signal y N (n ) = y N( a ) (n(1 + δ )T ) is then
written

y N (n ) = ∑d
m ∈Z
N ,n −m g ( a ) (mT + nδT ) + v(n ) (8.28)

where v(n ) = v ( a ) (n(1 + δ )T ) is assumed white Gaussian circular with

zero mean and known variance σ 2 = E [| v(n )|2 ]. As usual, g(a)(lT) is
assumed time limited with the time support included in [0, LT] where
L is a known integer. We thus write gl = g(a)(lT) for l = 0, ..., L−1.
13 0 Im ag e En c ry p ti o n

65
FFT+clock offset
60
DCT+clock offset
DWT+clock offset
55
FFT+clock offset+ZP
50 DCT+clock offset+ZP
DWT+clock offset+ZP
45
PSNR(dB)

40

35

30

25

20

15
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 8.22  PSNR versus Eb /N0 for chaotic-encrypted image transmission with OFDM system
over a Raleigh fading channel.

45
FFT + clock offset
DCT + clock offset
40
DWT + clock offset
FFT + clock offset + ZP
DCT + clock offset + ZP
35
DWT + clock offset + ZP
PSNR (dB)

30

25

20

15
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 8.23  PSNR versus Eb /N0 for chaotic-OFB-encrypted image transmission with OFDM
system over a Raleigh fading channel.
 OFDM Limitati o ns 131

The performance of the process of original image ­transmission

with the chaotic and chaotic-OFB algorithm for (FFT/DCT/
DWT)-OFDM systems over a Raleigh fading channel is shown in
Figures 8.22 and 8.23. From these figures, it is clear that the PSNR
performance of both chaotic and chaotic-OFB-encrypted image
transmission with FFT/DCT/DWT-OFDM is better with applying
the zero padding (ZP) guard and PSE method. At high Eb /N0 values,
it is possible to receive the chaotic-OFB-encrypted images with a very
large PSNR if error-­correction codes are utilized. From all obtained
results, we can see that sampling clock offset reduction technique
effects can be mitigated to a great extent. Another conclusion is that
the chaotic-OFB encryption preserves its rank as the best encryption
algorithm from the immunity to noise point of view even in the pres-
ence of sampling clock offset.
 9
Simulation Examples

9.1  Simulation Parameters

In our simulations, we carried out the encryption of the Cameraman

image with the previously discussed encryption algorithms. Then, the
encrypted images were transformed into a binary format and used for
orthogonal frequency division multiplexing (OFDM) modulation. Both
additive white Gaussian noise (AWGN) and Rayleigh fading channels
were considered in the simulations. A normalized carrier frequency
offset (CFO) of 0.1 was used in some of the simulation experiments.
PAPR reduction techniques were used in some other experiments. In all
experiments, 128 subcarriers and quaternary phase shift keying (QPSK)
were used. The simulation parameters are summarized in Table 9.1.
To evaluate the quality of the decrypted images at the receiver,
we used the peak signal-to-noise ratio (PSNR) between the o­ riginal
image and the decrypted image. Figures  9.1 to 9.3 illustrate the
­variation of the PSNR of the decrypted image with the signal-to-
noise ratio (SNR) in the channel (Eb /N0) for the fast Fourier t­ ransform
(FFT)-OFDM s­ystem with different encryption algorithms in an
AWGN channel for different numbers of subcarriers. These figures
showed that as the number of subcarriers is increased, the quality of
the decrypted images is improved.
In this section, several experiments compare between the encryp-
tion algorithms and select the most suitable system for the transmission
of encrypted images. To better illustrate the results, each encryp-
tion algorithm was studied for a different number of ­subchannels
as shown in Figures 9.4 to 9.6 without equalization and Figures 9.7
to 9.9 with zero-forcing (ZF) equalization. From these figures, it is
clear that equalization enhances the system performance dramati-
cally. The best-obtained results from an encryption algorithm were
for the c­ haotic encryption algorithm because it is a permutation-based
ciphering algorithm.

13 3
13 4 Im ag e En c ry p ti o n

Table 9.1  Simulation Parameters

PARAMETER DESCRIPTION
Transmitter System bandwidth 64 MHz
Modulation type QPSK
CP length 16 samples
N 128 subcarriers
Subcarrier spacing 500 kHz
Type of transform FFT, DCT, or DWT
Channel Channel model AWGN or Rayleigh fading
Receiver PSE and equalizer ZF equalization
Encryption algorithm Type of encryption Chaotic encryption, DES, AES, or RC6
Key of DES (00010203040506)16
Key of RC6 (000102030405060708090a0b0c0d0e0f)16
Key of AES (000102030405060708090a0b0c0d0e0f)16

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption
50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.1  PSNR vs. Eb /N0 in the FFT-OFDM system for an AWGN channel with N = 64.

The Data Encryption Standard (DES), the Advanced Encryption

Standard (AES), and the RC6 encryption algorithms provided
lower PSNR performance for the FFT-OFDM system compared
with the chaotic algorithm because these algorithms have a diffu-
sion ­mechanism, which reduces the noise immunity. On the other
hand, the chaotic algorithm is more robust to noise. From these
 Simul ati o n E x a m p l e s 13 5

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.2  PSNR versus Eb /N0 in the FFT-OFDM system for an AWGN channel with N = 512.

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption
50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.3  PSNR versus Eb /N0 in the FFT-OFDM system for an AWGN channel with N = 1024.
13 6 Im ag e En c ry p ti o n

9.3

9.2

9.1

8.9
PSNR(dB)

8.8

8.7
Original image
8.6
Chaotic image encryption
8.5 DES image encryption
AES image encryption
8.4 RC6 image encryption

8.3
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.4  PSNR versus Eb /N0 in the FFT-OFDM system for a Raleigh channel with N = 64 before
PSE and equalization.

11

10.5

Original image
10 Chaotic image encryption
DES image encryption
PSNR(dB)

AES image encryption

9.5 RC6 image encryption

8.5

8
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.5  PSNR versus Eb /N0 in the FFT-OFDM system for a Raleigh channel with N = 512
before PSE and equalization.
 Simul ati o n E x a m p l e s 13 7

11

10.5

10
PSNR(dB)

9.5

9
Original image
Chaotic image encryption
8.5 DES image encryption
AES image encryption
RC6 image encryption
8
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.6  PSNR versus Eb /N0 in the FFT-OFDM system for a Raleigh channel with N = 1024
before PSE and equalization.

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.7  PSNR versus Eb /N0 in the FFT-OFDM system for a Raleigh channel with N = 64 after
PSE and equalization.
13 8 Im ag e En c ry p ti o n

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.8  PSNR versus Eb /N0 in the FFT-OFDM system for a Raleigh channel with N = 512 after
PSE and equalization.

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.9  PSNR versus Eb /N0 in the FFT-OFDM system for a Raleigh channel with N = 1024
after PSE and equalization.
 Simul ati o n E x a m p l e s 13 9

figures, it  is clear that equalization enhances the system perfor-

mance d ­ ramatically. The best-obtained results from an encryption
algorithm were for the chaotic encryption algorithm because it is a
permutation-based ciphering algorithm.
The effect of the guard interval length on the system was s­ tudied;
Figures 9.10 to 9.12 illustrate the PSNR performance versus the Eb /N 0
80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.10  PSNR versus Eb /N0 in the FFT-OFDM system for an AWGN channel with G = 0.

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.11  PSNR versus Eb /N0 in the FFT-OFDM system for an AWGN channel with G = 16.
14 0 Im ag e En c ry p ti o n

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.12  PSNR versus Eb /N0 in the FFT-OFDM system for an AWGN channel with G = 32.

for the FFT-OFDM systems with different encryption ­a lgorithms in

an AWGN channel. The guard intervals utilized ranged from 0 to 32
samples. From these figures, it is clear that the PSNR p­ erformances
of all encryption algorithms with the FFT-OFDM system at G = 0
were the best compared with those with G = 16 and 32. The chaotic-
encryption gives the best PSNR values.
Figures  9.13 to 9.15 illustrate the variation of the PSNR with
the Eb /N 0 for the FFT-OFDM with G = [0, 16, 32], with differ-
ent encryption algorithms in a frequency-selective channel. Without
pilot symbol estimation (PSE), it was shown that fading disrupts
the PSNR performance of FFT-OFDM systems regardless of the
encryption algorithm. However, Figures 9.16 to 9.18 show that, with
PSE, the PSNR performances of all systems were better than those
without PSE.
The DES, AES, and RC6 provide lower PSNR performance for
the FFT-OFDM system than the chaotic algorithm because these
algorithms have a diffusion mechanism, which highly reduces the
noise immunity. On the other hand, the chaotic algorithm is more
robust to noise.
 Simul ati o n E x a m p l e s 141

11

Original image
10.5 Chaotic image encryption
DES image encryption
AES image encryption
10 RC6 image encryption
PSNR(dB)

9.5

8.5

8
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.13  PSNR versus Eb /N0 in the FFT-OFDM system for a Raleigh channel with G = 0 with-
out PSE and equalization.

11

Original image
10.5 Chaotic image encryption
DES image encryption
AES image encryption
10 RC6 image encryption
PSNR(dB)

9.5

8.5

8
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.14  PSNR versus Eb /N0 in the FFT-OFDM system for a Raleigh channel with G = 16
without PSE and equalization.
14 2 Im ag e En c ry p ti o n

11

Original image
10.5 Chaotic image encryption
DES image encryption
AES image encryption
10 RC6 image encryption
PSNR(dB)

9.5

8.5

8
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.15  PSNR versus Eb /N0 in the FFT-OFDM system for a Raleigh channel with G = 32
without PSE and equalization.

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.16  PSNR versus Eb /N0 in the FFT-OFDM system for a Raleigh channel with G = 0 with
PSE and equalization.
 Simul ati o n E x a m p l e s 14 3

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.17  PSNR versus Eb /N0 in the FFT-OFDM system for a Raleigh channel with G = 16 with
PSE and equalization.

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption
50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.18  PSNR versus Eb /N0 in the FFT-OFDM system for a Raleigh channel with G = 32 with
PSE and equalization.
14 4 Im ag e En c ry p ti o n

The different guard intervals were studied. A comparison study

between the hybrid guard interval (HGI), which is composed of a cyclic
prefix (CP) and zero samples, and the zero padding (ZP) guard inter-
val with the different encryption algorithms in a frequency-selective
channel is illustrated in Figures 9.19 and 9.20. CFO compensation was
performed. It is shown that the PSNR performance of the ZP-OFDM
systems is slightly better than that of the HGI-OFDM systems.
Figure  9.21 gives the output of the FFT-OFDM system for the
transmission of the original Cameraman image over an AWGN
channel at N = 128 and G = 32.
Figure  9.22 gives the output of the FFT-OFDM system for the
transmission of the original Cameraman image over a fading channel
at N = 128 and G = 32.
Figure  9.23 gives the output of the FFT-OFDM system for the
transmission of the chaotic-encrypted image over an AWGN channel
at N = 128 and G = 32.
Figure  9.24 gives the output of the FFT-OFDM system for the
transmission of the chaotic-encrypted image over a fading channel at
N = 128 and G = 32.

9.3

9.2

9.1

9 Original with HGI

Chaotic with HGI
8.9 DES with HGI
PSNR(dB)

AES with HGI

RC6 with HGI
8.8 Original with ZP
Chaotic with ZP
DES with ZP
8.7 AES with ZP
RC6 with ZP
8.6

8.5

8.4

8.3
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.19  PSNR versus Eb /N0 in the FFT-OFDM system with ZP and hybrid G for a Raleigh
channel with N = 128 and G = 32 without PSE and equalization.
 Simul ati o n E x a m p l e s 14 5

80
Original with HGI
70 Chaotic with HGI
DES with HGI
AES with HGI
60
RC6 with HGI
Original with ZP
50 Chaotic with ZP
DES with ZP
PSNR(dB)

AES with ZP
40 RC6 with ZP

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.20  PSNR versus Eb /N0 in the FFT-OFDM system with ZP and HGI for a Raleigh channel
with N = 128 and G = 32 with PSE and equalization.

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 14.6993 PSNR = 23.7838

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 48.0593 PSNR >60

Figure 9.21  Original image with FFT-OFDM over an AWGN channel.

14 6 Im ag e En c ry p ti o n

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 17.8214 PSNR = 29.4547

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 53.8719 PSNR >60

Figure 9.22  Original image with FFT-OFDM over a fading channel.

Figure  9.25 gives the output of the FFT-OFDM system for the
transmission of the DES-encrypted image over an AWGN channel
at N = 128 and G = 32.
Figure  9.26 gives the output of the FFT-OFDM system for the
transmission of the DES-encrypted image over a fading channel at
N = 128 and G = 32.
Figure  9.27 gives the output of the FFT-OFDM system for the
transmission of the AES-encrypted image over an AWGN channel
at N = 128 and G = 32.
Figure  9.28 gives the output of the FFT-OFDM system for the
transmission of the AES-encrypted image over a fading channel at
N = 128 and G = 32.
 Simul ati o n E x a m p l e s 147

(a)Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 15.1527 PSNR = 25.1709

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 50.5529 PSNR >60

Figure 9.23  Chaotic-encrypted image with FFT-OFDM over an AWGN channel.

Figure  9.29 gives the output of the FFT-OFDM system for the
transmission of the RC6-encrypted image over an AWGN channel
at N = 128 and G = 32.
Figure  9.30 gives the output of the FFT-OFDM system for the
transmission of the RC6-encrypted image over a fading channel at
N = 128 and G = 32.
Figures  9.31 to 9.33 illustrate the variation of the PSNR of a
decrypted image with the SNR in the channel (Eb /N0) for the DCT-
OFDM system with different encryption algorithms in an AWGN
channel for different numbers of subcarriers. These figures show that
as the number of subcarriers is increased, the quality of the decrypted
images improved.
14 8 Im ag e En c ry p ti o n

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 18.3597 PSNR = 30.0828

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 53.8881 PSNR >60

Figure 9.24  Chaotic-encrypted image with FFT-OFDM over a fading channel.

To better illustrate the results, each encryption algorithm was

­studied for a different number of subchannels as shown in Figures 9.34
to 9.36 without equalization and Figures  9.37 to 9.39 with ZF
­equalization. From these figures, it is clear that equalization enhances
the system performance dramatically. The best-obtained results from
an encryption algorithm were for the chaotic encryption algorithm
because it is a permutation-based ciphering algorithm.
The effect of the guard interval length on the system is studied in
Figures 9.40 to 9.42, which illustrate the PSNR performance versus
the Eb /N0 for the DCT-OFDM systems with different encryption
algorithms in an AWGN channel. The guard intervals utilized range
from 0 to 32 samples. From these figures, it is clear that the PSNR
performances of all encryption algorithms with the DCT-OFDM
 Simul ati o n E x a m p l e s 14 9

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 8.3766 PSNR = 11.1743

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 36.4777 PSNR >60

Figure 9.25  DES-encrypted image with FFT-OFDM over an AWGN channel.

system at G = 0 are the best compared with those with G = 16 and 32.
The chaotic-encrypted image gives the best PSNR values.
Figures  9.43 to 9.45 illustrate the variation of the PSNR with
the Eb /N0 for the DCT-OFDM with G = [0, 16, 32], with different
­encryption algorithms in a frequency-selective channel. Without PSE,
it is shown that fading disrupts the PSNR performance of the DCT-
OFDM system, regardless of the encryption algorithm. However,
Figures 9.46 to 9.48 show that, with PSE, the PSNR ­performances of
all systems were better than those without PSE.
The different types of guard intervals were studied. A comparison
study between the HGI, which is composed of a CP and zero samples,
and the ZP guard interval with the different encryption algorithms in
a frequency-selective channel is illustrated in Figures 9.49 and 9.50.
15 0 Im ag e En c ry p ti o n

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 9.6637 PSNR = 16.7520

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 41.4821 PSNR >60

Figure 9.26  DES-encrypted image with FFT-OFDM over a fading channel.

CFO compensation was performed. It is shown that the PSNR

­performance of the ZP-OFDM system was slightly better than that
of the HGI-OFDM system.
Figure 9.51 gives the output of the DCT-OFDM system for the
transmission of the original Cameraman image over an AWGN
channel at N = 128 and G = 32.
Figure 9.52 gives the output of the DCT-OFDM system for the
transmission of the original Cameraman image over a fading channel
at N = 128 and G = 32.
Figure 9.53 gives the output of the DCT-OFDM system for the
transmission of the chaotic-encrypted image over an AWGN channel
at N = 128 and G = 32.
 Simul ati o n E x a m p l e s 151

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 8.4026 PSNR = 9.4551

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 31.5464 PSNR >60

Figure 9.27  AES-encrypted image with FFT-OFDM over an AWGN channel.

Figure 9.54 gives the output of the DCT-OFDM system for the

transmission of the chaotic-encrypted image over a fading channel at
N = 128 and G = 32.
Figure 9.55 gives the output of the DCT-OFDM system for the
transmission of the DES-encrypted image over an AWGN channel
at N = 128 and G = 32.
Figure 9.56 gives the output of the DCT-OFDM system for the
transmission of the DES-encrypted image over a fading channel at
N = 128 and G = 32.
Figure 9.57 gives the output of the DCT-OFDM system for the
transmission of the AES-encrypted image over an AWGN channel
at N = 128 and G = 32.
15 2 Im ag e En c ry p ti o n

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 8.6932 PSNR = 14.1673

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 36.8018 PSNR >60

Figure 9.28  AES-encrypted image with FFT-OFDM over a fading channel.

Figure 9.58 gives the output of the DCT-OFDM system for the

transmission of the AES-encrypted image over a fading channel at
N = 128 and G = 32.
Figure 9.59 gives the output of the DCT-OFDM system for the
transmission of the RC6-encrypted image over an AWGN channel at
N = 128 and G = 32.
Figure 9.60 gives the output of the DCT-OFDM system for the
transmission of the RC6-encrypted image over a fading channel at
N = 128 and G = 32.
Figures  9.61 to 9.63 illustrate the variation of the PSNR of a
decrypted image with the SNR in the channel (Eb /N0) for the DWT-
OFDM system with different encryption algorithms in an AWGN
channel for different numbers of subcarriers. These figures show that
 Simul ati o n E x a m p l e s 15 3

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 8.3916 PSNR = 9.4165

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 30.9051 PSNR >60

Figure 9.29  RC6-encrypted image with FFT-OFDM over an AWGN channel.

as the number of subcarriers is increased, the quality of the decrypted

images is improved.
Each encryption algorithm was studied for a different number of
subchannels as shown in Figures  9.64 to 9.66 without equalization
and Figures 9.67 to 9.69 with ZF equalization. From these figures, it
is clear, as obtained in the previous results, that equalization enhanced
system performance. The best-obtained results from an encryption
algorithm were for the chaotic encryption algorithm because it is a
permutation-based ciphering algorithm. The DES, the AES, and
the RC6 encryption algorithms provides worse PSNR performance
for the DWT-OFDM system compared with the chaotic algorithm
because these algorithms have a diffusion mechanism that reduces the
15 4 Im ag e En c ry p ti o n

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 8.7540 PSNR = 13.8893

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 40.0734 PSNR >60

Figure 9.30  RC6-encrypted image with FFT-OFDM over a fading channel.

noise immunity. On the other hand, the chaotic algorithm is more

robust to noise.
The effect of the guard interval length on the system was studied;
Figures 9.70 to 9.72 illustrate the PSNR performance versus the Eb /N0
for the DWT-OFDM systems with different encryption ­algorithms in
an AWGN channel. The guard intervals utilized ranged from 0 to 32
samples. From these figures, it is clear that the PSNR ­performances
of all encryption algorithms with the DWT-OFDM system at G = 0
were the best compared with those with G = 16 and 32. The chaotic
encryption gives the best PSNR values.
Figures  9.73 to 9.75 illustrate the variation of the PSNR with
the Eb /N0 for the DWT-OFDM with G = [0, 16, 32], with differ-
ent encryption algorithms in a frequency-selective channel. Without
 Simul ati o n E x a m p l e s 15 5

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.31  PSNR versus Eb /N0 in the DCT-OFDM system for an AWGN channel with N = 64.

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.32  PSNR versus Eb /N0 in the DCT-OFDM system for an AWGN channel with N = 512.
15 6 Im ag e En c ry p ti o n

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.33  PSNR versus Eb /N0 in the DCT-OFDM system for an AWGN channel with N = 1024.

11

Original image
10.5 Chaotic image encryption
DES image encryption
AES image encryption
10 RC6 image encryption
PSNR(dB)

9.5

8.5

8
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.34  PSNR versus Eb /N0 in the DCT-OFDM system for a Raleigh channel with N = 64
without PSE and equalization.
 Simul ati o n E x a m p l e s 15 7

11
Original image
Chaotic image encryption
10.5 DES image encryption
AES image encryption
RC6 image encryption
10
PSNR(dB)

9.5

8.5

8
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.35  PSNR versus Eb /N0 in the DCT-OFDM system for a Raleigh channel with N = 512
without PSE and equalization.

11
Original image
Chaotic image encryption
10.5 DES image encryption
AES image encryption
RC6 image encryption
10
PSNR(dB)

9.5

8.5

8
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.36  PSNR versus Eb /N0 in the DCT-OFDM system for a Raleigh channel with N = 1024
without PSE and equalization.
15 8 Im ag e En c ry p ti o n

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption
50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.37  PSNR versus Eb /N0 in the DCT-OFDM system for a Raleigh channel with N = 64 with
PSE and equalization.

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.38  PSNR versus Eb /N0 in the DCT-OFDM system for a Raleigh channel with N = 512
with PSE and equalization.
 Simul ati o n E x a m p l e s 15 9

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.39  PSNR versus Eb /N0 in the DCT-OFDM system for a Raleigh channel with N = 1024
with PSE and equalization.

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.40  PSNR versus Eb /N0 in the DCT-OFDM system for an AWGN channel with G = 0.
16 0 Im ag e En c ry p ti o n

80
Original image
70 Chaotic image encryption
DES image encryption
AES image encryption
60
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.41  PSNR versus Eb /N0 in the DCT-OFDM system for an AWGN channel with G = 16.

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.42  PSNR versus Eb /N0 in the DCT-OFDM system for an AWGN channel with G = 32.
 Simul ati o n E x a m p l e s 161

11
Original image
Chaotic image encryption
10.5 DES image encryption
AES image encryption
RC6 image encryption
10
PSNR(dB)

9.5

8.5

8
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.43  PSNR versus Eb /N0 in the DCT-OFDM system for a Raleigh channel with G = 0
without PSE and equalization.

11
Original image
Chaotic image encryption
10.5 DES image encryption
AES image encryption
RC6 image encryption
10
PSNR(dB)

9.5

8.5

8
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.44  PSNR versus Eb /N0 in the DCT-OFDM system for a Raleigh channel with G = 16
without PSE and equalization.
16 2 Im ag e En c ry p ti o n

11
Original image
Chaotic image encryption
10.5 DES image encryption
AES image encryption
RC6 image encryption
10
PSNR(dB)

9.5

8.5

8
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.45  PSNR versus Eb /N0 in the DCT-OFDM system for a Raleigh channel with G = 32
without PSE and equalization.

80
Original image
70 Chaotic image encryption
DES image encryption
AES image encryption
60
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.46  PSNR versus Eb /N0 in the DCT-OFDM system for a Raleigh channel with G = 0 with
PSE and equalization.
 Simul ati o n E x a m p l e s 16 3

80
Original image
70 Chaotic image encryption
DES image encryption
AES image encryption
60
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.47  PSNR versus Eb /N0 in the DCT-OFDM system for a Raleigh channel with G = 16 with
PSE and equalization.

80
Original image
70 Chaotic image encryption
DES image encryption
AES image encryption
60
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.48  PSNR versus Eb /N0 in the DCT-OFDM system for a Raleigh channel with G = 32 with
PSE and equalization.
16 4 Im ag e En c ry p ti o n

9.3

9.2

9.1

8.9
PSNR(dB)

8.8
Original with HGI
8.7 Chaotic with HGI
DES with HGI
8.6 AES with HGI
RC6 with HGI
8.5 Original with ZP
Chaotic with ZP
DES with ZP
8.4 AES with ZP
RC6 with ZP
8.3
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.49  PSNR versus Eb /N0 in the DCT-OFDM system with ZP and HGI for a Raleigh channel
with N = 128 and G = 32 without PSE and equalization.

80

Original with HGI

70 Chaotic with HGI
DES with HGI
60 AES with HGI
RC6 with HGI
Original with ZP
50 Chaotic with ZP
DES with ZP
PSNR(dB)

AES with ZP
40
RC6 with ZP

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.50  PSNR versus Eb /N0 in the DCT-OFDM system with ZP and HGI for a Raleigh channel
with N = 128 and G = 32 with PSE and equalization.
 Simul ati o n E x a m p l e s 16 5

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 15.7172 PSNR = 26.3188

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 53.5993 PSNR >60

Figure 9.51  Original image with DCT-OFDM over an AWGN channel.

PSE and equalization, it is shown that fading disrupts the PSNR per-
formance of the DWT-OFDM systems regardless of the encryption
algorithm. However, Figures 9.76 to 9.78 show that, with PSE and
equalization, the PSNR performances of all systems were better than
those without PSE and equalization.
The DES, AES, and RC6 provide lower PSNR performance for
the DWT-OFDM system than the chaotic algorithm because these
algorithms have a diffusion mechanism that highly reduces the noise
immunity. On the other hand, the chaotic algorithm was more robust
to noise.
The different types of guard intervals were studied. A comparison
study between the HGI and the ZP guard interval with the different
encryption algorithms in a frequency-selective channel is illustrated
16 6 Im ag e En c ry p ti o n

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 16.7843 PSNR = 23.2367

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 27.9452 PSNR = 29.3708

Figure 9.52  Original image with DCT-OFDM over a fading channel.

in Figures  9.79 and 9.80. CFO compensation was performed. The

PSNR performance of the ZP-OFDM systems was slightly better
than that of the HGI-OFDM systems.
Figure 9.81 gives the output of the DWT-OFDM system for the
transmission of the original Cameraman image over an AWGN
channel at N = 128 and G = 32.
Figure 9.82 gives the output of the DWT-OFDM system for the
transmission of the original Cameraman image over a fading channel
at N = 128 and G = 32.
Figure 9.83 gives the output of the DWT-OFDM system for the
transmission of the chaotic-encrypted image over an AWGN channel
at N = 128 and G = 32.
 Simul ati o n E x a m p l e s 16 7

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 15.0481 PSNR = 24.4613

(c) Eb/N0 = 10 dB (d) Eb/N0= 15 dB

PSNR = 53.4971 PSNR >60

Figure 9.53  Chaotic-encrypted image with DCT-OFDM over an AWGN channel.

Figure 9.84 gives the output of the DWT-OFDM system for the

transmission of the chaotic-encrypted image over a fading channel at
N = 128 and G = 32.
Figure 9.85 gives the output of the DWT-OFDM system for the
transmission of the DES-encrypted image over an AWGN channel
at N = 128 and G = 32.
Figure 9.86 gives the output of the DWT-OFDM system for the
transmission of the DES-encrypted image over a fading channel at
N = 128 and G = 32.
Figure 9.87 gives the output of the DWT-OFDM system for the
transmission of the AES-encrypted image over an AWGN channel
at N = 128 and G = 32.
16 8 Im ag e En c ry p ti o n

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 16.3858 PSNR = 23.0375

(c) Eb/N0 = 10 dB (d) Eb/N0= 15 dB

PSNR = 30.3995 PSNR = 33.2591

Figure 9.54  Chaotic-encrypted image with DCT-OFDM over a fading channel.

Figure 9.88 gives the output of the DWT-OFDM system for the

transmission of the AES-encrypted image over a fading channel at
N = 128 and G = 32.
Figure 9.89 gives the output of the DWT-OFDM system for the
transmission of the RC6-encrypted image over an AWGN channel at
N = 128 and G = 32.
Figure 9.90 gives the output of the DWT-OFDM system for the
transmission of the RC6-encrypted image over a fading channel at
N = 128 and G = 32.
Table 9.2 summarizes these results.
 Simul ati o n E x a m p l e s 16 9

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 8.3762 PSNR = 11.1133

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 34.4818 PSNR >60

Figure 9.55  DES-encrypted image with DCT-OFDM over an AWGN channel.

9.2  Simulation Experiments in the Presence of CFO

Figure 9.91 illustrates the variation of the PSNR with the Eb/N0 for
the FFT-OFDM, DCT-OFDM, and DWT-OFDM s­ystems with
different encryption algorithms. A symbol rate of 250,000 ­symbols
per second, ε = 0.1, N = 128, and AWGN channel are assumed. This
figure shows that the PSNR performance of all systems with different
encryption algorithms deteriorates due to the presence of the CFO
because the CFO disrupts the orthogonality between subcarriers and
gives rise to the intercarrier interference (ICI), which leads to perfor-
mance degradation. This figure also shows that the CFO compensa-
tion process can avoid the impact of the CFO and produce high-quality
17 0 Im ag e En c ry p ti o n

(a) Eb/N0= 0 dB (b) Eb/N0 = 5 dB

PSNR = 9.0390 PSNR = 11.7067

(c) Eb/N0= 10 dB (d) Eb/N0= 15 dB

PSNR = 16.5200 PSNR = 19.9308

Figure 9.56  DES-encrypted image with DCT-OFDM over a fading channel.

decrypted images. On the other hand, without CFO compensation, it

is clear that the CFO degrades the PSNR p ­ erformance of the DES,
AES, and RC6 for all OFDM systems.
Figures  9.92 to 9.97 illustrate the PSNR performance versus the
Eb /N0 for the FFT-OFDM, DCT-OFDM, and DWT-OFDM sys-
tems with different encryption algorithms in a frequency-selective
channel. Without CFO compensation, it is shown that CFO dis-
rupted the PSNR performance of all OFDM systems regardless of
the encryption algorithm. However, with CFO compensation, the
PSNR performances of all systems are better than those without
CFO compensation. These figures also show that the PSNR perfor-
mances of all encryption algorithms with the FFT-OFDM system
 Simul ati o n E x a m p l e s 171

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 8.4211 PSNR = 9.4006

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 30.5205 PSNR >60

Figure 9.57  AES-encrypted image with DCT-OFDM over an AWGN channel.

were the best compared with those of the DCT-OFDM and DWT-
OFDM systems.
The DES, AES, and RC6 provide lower PSNR p ­ erformance for
all OFDM systems compared with the chaotic algorithm because
these algorithms have a diffusion mechanism that reduces the noise
immunity. On the other hand, the chaotic algorithm is more robust
to noise.

9.3  Simulation Experiments for Enhanced Algorithms

The performance of the chaotic algorithm with its modes (ECB,

CFB, CBC, and OFB) for (FFT/DCT/DWT)-OFDM systems was
compared. With N = 128 and G = 32 over an AWGN, the PSNR
17 2 Im ag e En c ry p ti o n

(a) Eb/N0= 0 dB (b) Eb/N0 = 5 dB

PSNR = 8.5829 PSNR = 10.2435

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 14.1012 PSNR = 16.2472

Figure 9.58  AES-encrypted image with DCT-OFDM over a fading channel.

performance of the chaotic-ECB, chaotic-CFB, chaotic-CBC, and

chaotic-OFB algorithms for all OFDM systems start to increase at
moderate-to-high Eb /N0 values. In Figure  9.98, the PSNR of the
chaotic-ECB algorithm is also the best compared to those of all
other modes of encryption algorithms with FFT-OFDM. From
Figure  9.99, the PSNR of the chaotic-(OFB and ECB) algorithms
is also the best compared to those of all other modes of encryption
algorithms with DCT-OFDM. From Figure  9.100, the PSNR of
the chaotic-OFB algorithm is the best compared to that of all other
modes of encryption algorithms with DWT-OFDM.
The performance of the chaotic algorithm with its modes (ECB,
CFB, CBC, and OFB) for (FFT/DCT/DWT)-OFDM systems
was compared. With N = 128 and G = 32 over a fading channel,
 Simul ati o n E x a m p l e s 173

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 8.3961 PSNR = 9.3528

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 33.1261 PSNR >60

Figure 9.59  RC6-encrypted image with DCT-OFDM over an AWGN channel.

the PSNR performance of the chaotic-ECB, chaotic-CFB, chaotic-

CBC, and chaotic-OFB algorithms for all OFDM systems start
to increase at moderate-to-high Eb /N 0 values. In Figures  9.101 to
9.103, the PSNR of the chaotic-CBC algorithm is the best compared
to those of all other modes of encryption with FFT/DCT/DWT-
OFDM without CFO. From Figures 9.104 to 9.106, the PSNR of
the chaotic-CBC algorithm is also the best compared to those of all
other modes of encryption algorithms with FFT-OFDM systems
with CFO.
A comparison study between the CP-OFDM systems and the
ZP-OFDM systems with the different chaotic encryption modes
in a frequency-selective channel without PSE is illustrated in
174 Im ag e En c ry p ti o n

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 8.5691 PSNR = 10.2276

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 14.1490 PSNR = 16.1107

Figure 9.60  RC6-encrypted image with DCT-OFDM over a fading channel.

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.61  PSNR versus Eb /N0 in the DWT-OFDM system for an AWGN channel with N = 64.
 Simul ati o n E x a m p l e s 17 5

80
Original image
70 Chaotic image encryption
DES image encryption
AES image encryption
60
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.62  PSNR versus Eb /N0 in the DWT-OFDM system for an AWGN channel with N = 512.

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.63  PSNR versus Eb /N0 in the DWT-OFDM system for an AWGN channel with N = 1024.
17 6 Im ag e En c ry p ti o n

11

Original image
10.5 Chaotic image encryption
DES image encryption
AES image encryption
10 RC6 image encryption
PSNR(dB)

9.5

8.5

8
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.64  PSNR versus Eb /N0 in the DWT-OFDM system for a Raleigh channel with N = 64
without PSE and equalization.

11

10.5

10 Original image
Chaotic image encryption
PSNR(dB)

DES image encryption

9.5 AES image encryption
RC6 image encryption

8.5

8
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.65  PSNR versus Eb /N0 in the DWT-OFDM system for a Raleigh channel with N = 512
without PSE and equalization.
 Simul ati o n E x a m p l e s 17 7

11

10.5

10
PSNR(dB)

9.5

9
Original image
Chaotic image encryption
8.5 DES image encryption
AES image encryption
RC6 image encryption
8
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.66  PSNR versus Eb /N0 in the DWT-OFDM system for a Raleigh channel with N = 1024
without PSE and equalization.

80
Original image
70 Chaotic image encryption
DES image encryption
AES image encryption
60 RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.67  PSNR versus Eb /N0 in the DWT-OFDM system for a Raleigh channel with N = 64
with PSE and equalization.
178 Im ag e En c ry p ti o n

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.68  PSNR versus Eb /N0 in the DWT-OFDM system for a Raleigh channel with N = 512
with PSE and equalization.

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.69  PSNR versus Eb /N0 in the DWT-OFDM system for a Raleigh channel with N = 1024
with PSE and equalization.
 Simul ati o n E x a m p l e s 179

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.70  PSNR versus Eb /N0 in the DWT-OFDM system for an AWGN channel with G = 0.

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.71  PSNR versus Eb /N0 in the DWT-OFDM system for an AWGN channel with G = 16.
18 0 Im ag e En c ry p ti o n

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption
50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.72  PSNR versus Eb /N0 in the DWT-OFDM system for an AWGN channel with G = 32.

11

Original image
10.5
Chaotic image encryption
DES image encryption
AES image encryption
10
RC6 image encryption
PSNR(dB)

9.5

8.5

8
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.73  PSNR versus Eb /N0 in the DWT-OFDM system for a Raleigh channel with G = 0
without PSE and equalization.
 Simul ati o n E x a m p l e s 181

11
Original image
Chaotic image encryption
10.5 DES image encryption
AES image encryption
RC6 image encryption
10
PSNR(dB)

9.5

8.5

8
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.74  PSNR versus Eb /N0 in the DWT-OFDM system for a Raleigh channel with G = 16
without PSE and equalization.

11
Original image
Chaotic image encryption
10.5 DES image encryption
AES image encryption
RC6 image encryption
10
PSNR(dB)

9.5

8.5

8
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.75  PSNR versus Eb /N0 in the DWT-OFDM system for a Raleigh channel with G = 32
without PSE and equalization.
18 2 Im ag e En c ry p ti o n

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.76  PSNR versus Eb /N0 in the DWT-OFDM system for a Raleigh channel with G = 0 with
PSE and equalization.

80
Original image
70 Chaotic image encryption
DES image encryption
60 AES image encryption
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.77  PSNR versus Eb /N0 in the DWT-OFDM system for a Raleigh channel with G = 16 with
PSE and equalization.
 Simul ati o n E x a m p l e s 18 3

80
Original image

70 Chaotic image encryption

DES image encryption
AES image encryption
60
RC6 image encryption

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.78  PSNR versus Eb /N0 in the DWT-OFDM system for a Raleigh channel with G = 32 with
PSE and equalization.

11
Original with HIG
Chaotic with HIG
10.5 DES with HIG
AES with HIG
RC6 with HIG
Original with ZP
10
Chaotic with ZP
DES with ZP
PSNR(dB)

AES with ZP
9.5 RC6 with ZP

8.5

8
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.79  PSNR versus Eb /N0 in the DWT-OFDM system with ZP and HGI for a Raleigh channel
with N = 128 and G = 32 without PSE and equalization.
18 4 Im ag e En c ry p ti o n

80
Original with HIG
Chaotic with HIG
70
DES with HIG
AES with HIG
60 RC6 with HIG
Original with ZP
Chaotic with ZP
50
DES with ZP
PSNR(dB)

AES with ZP
40 RC6 with ZP

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.80  PSNR versus Eb /N0 in the DWT-OFDM system with ZP and HGI for a Raleigh channel
with N = 128 and G = 32 with PSE and equalization.

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 14.6839 PSNR = 23.7368

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 45.9748 PSNR >60

Figure 9.81  Original image with DWT-OFDM over an AWGN channel.

 Simul ati o n E x a m p l e s 18 5

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 17.5731 PSNR = 24.9482

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 32.6435 PSNR = 37.7109

Figure 9.82  Original image with DWT-OFDM over a fading channel.

Figures  9.107 to 9.109. The best value of PSNR is obtained with

chaotic-ECB. In Figures 9.110 to Figure 9.112, they are considered.
It is shown that the PSNR performance of the ZP-OFDM systems
is better than that of the CP-OFDM systems. In Figures 9.113 to
9.115, it is illustrated that over an AWGN channel the PSNR per-
formance of the OFDM systems with CFO is better than that of
OFDM systems without CFO.
Figure 9.116 compares the AES encryption algorithm with AES
with preprocessing in an AWGN channel when N = 128 and G =
32 to find that FFT-OFDM is the best, then DCT-OFDM, and
finally DWT-OFDM. Figure 9.117 illustrates a fading channel case
without PSE and equalization; Figure 9.118 illustrates a fading chan-
nel also, but with PSE and equalization. CFO is considered without
18 6 Im ag e En c ry p ti o n

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 14.7090 PSNR = 23.5544

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 44.6552 PSNR >60

Figure 9.83  Chaotic-encrypted image with DWT-OFDM over an AWGN channel.

compensation in Figure 9.119; Figure 9.120 illustrates CFO compen-

sation. Figures  9.121 and 9.122 consider an AWGN channel with
CP-OFDM and ZP-OFDM, respectively.
Figure 9.123 compares the RC6 encryption algorithm with RC6
with preprocessing in an AWGN channel when N = 128 and G = 32;
FFT-OFDM is the best, then DCT-OFDM, and finally DWT-
OFDM. Figure 9.124 illustrates a fading channel case without PSE
equalization; Figure  9.125 illustrates a fading channel case also,
but with PSE and equalization. CFO is considered in Figure 9.126
without compensation; Figure  9.127 illustrates CFO compensa-
tion. Figures  9.128 and 9.129 consider an AWGN channel with
CP-OFDM and ZP-OFDM, respectively.
 Simul ati o n E x a m p l e s 18 7

(a) Eb/N0= 0 dB (b) Eb/N0 = 5 dB

PSNR = 16.3689 PSNR = 23.8108

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 32.7138 PSNR = 37.5838

Figure 9.84  Chaotic-encrypted image with DWT-OFDM over a fading channel.

9.4  Simulation Experiments of PAPR Reduction Methods

Several experiments were carried out to test the effect of the PAPR
reduction methods on the process of encrypted image transmission
with OFDM. A convolutional code with rate 1/2, constraint length
7, and octal generator polynomial (133,171) are adopted to reduce
the effect of PAPR reduction on the quality of received images. The
compared PAPR reduction methods are clipping, companding, and
hybrid clipping-and-companding methods.
The performance of the process of original image transmission with
the chaotic algorithm and three different diffusion algorithms for
(FFT/DCT/DWT)-OFDM systems with N = 128 and G = 32 over
an AWGN is shown in Figures 9.130 to 9.132. From these figures, it
18 8 Im ag e En c ry p ti o n

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 8.3954 PSNR = 11.2270

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 32.2373 PSNR >60

Figure 9.85  DES-encrypted image with DWT-OFDM over an AWGN channel.

is clear that the PSNR performance of the original image transmis-

sion process with FFT-OFDM is better with µ = 4 than with µ = 0.1;
for the DCT-OFDM and the DWT-OFDM, it is better with µ = 0.1
than with µ = 4.
Several experiments were repeated to study the effect of PAPR
reduction techniques with encrypted image transmission. From
Figures  9.133 to 9.135, the PSNR performance of the chaotic-
encrypted image transmission with FFT/DCT/DWT-OFDM was
better with µ = 0.1 than with µ = 4. At high SNR values, it is possible
to receive the chaotic encrypted images with a very large PSNR if
error correction codes were utilized.
 Simul ati o n E x a m p l e s 18 9

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 8.4729 PSNR = 10.7940

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 17.4856 PSNR = 25.9770

Figure 9.86  DES-encrypted image with DWT-OFDM over a fading channel.

Similar experiments were repeated for AES-encrypted images,

and the results are shown in Figures 9.136 to 9.138. We can come to
the conclusion that error-correcting codes are feasible, but the perfor-
mance is much less than that for the chaotic encryption case due to
the diffusion mechanism of the encryption algorithm.
Similar experiments were repeated for DES-encrypted images, and
the results are shown in Figures 9.139 to 9.141. Similar experiments
were repeated for RC6-encrypted images, and the results are shown
in Figures 9.142 to 9.144.
19 0 Im ag e En c ry p ti o n

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 8.4049 PSNR = 9.3407

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 28.8218 PSNR >60

Figure 9.87  AES-encrypted image with DWT-OFDM over an AWGN channel.

 Simul ati o n E x a m p l e s 191

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 8.3737 PSNR = 9.2508

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 13.7209 PSNR = 22.1366

Figure 9.88  AES-encrypted image with DWT-OFDM over a fading channel.

19 2 Im ag e En c ry p ti o n

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 8.4344 PSNR = 9.3419

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 29.1988 PSNR >60

Figure 9.89  RC6-encrypted image with DWT-OFDM over an AWGN channel.

 Simul ati o n E x a m p l e s 19 3

(a) Eb/N0 = 0 dB (b) Eb/N0 = 5 dB

PSNR = 8.3983 PSNR = 9.2497

(c) Eb/N0 = 10 dB (d) Eb/N0 = 15 dB

PSNR = 13.9828 PSNR = 18.7859

Figure 9.90  RC6-encrypted image with DWT-OFDM over a fading channel

19 4 Im ag e En c ry p ti o n

Table 9.2  Summarized Results

EB/N0 EB/N0 = 0 dB EB/N0 = 5 dB EB/N0 = 10 dB
ORIGINAL + AWGN CHANNEL
FFT 14.6993 23.7838 48.0593
DCT 15.7172 26.3188 53.5993
DWT 14.6839 23.7368 45.9748
ORIGINAL + FADING CHANNEL
FFT 17.8214 29.4547 53.8719
DCT 16.7843 23.2367 27.9452
DWT 17.5731 24.9482 32.6435
CHAOTIC + AWGN CHANNEL
FFT 15.1527 25.1709 50.5529
DCT 15.0481 24.4613 53.4971
DWT 14.7090 23.5544 44.6552
CHAOTIC + FADING CHANNEL
FFT 18.3597 30.0828 53.8881
DCT 16.3858 23.0375 30.3995
DWT 16.3689 23.8108 32.7138
DES + AWGN CHANNEL
FFT 8.3766 11.1743 36.4777
DCT 8.3762 11.1133 34.4818
DWT 8.3954 11.2270 32.2373
DES + FADING CHANNEL
FFT 9.6637 16.7520 41.4821
DCT 9.0390 11.7067 16.5200
DWT 8.4729 10.7940 17.4856
AES + AWGN CHANNEL
FFT 8.4026 9.4551 31.5464
DCT 8.4211 9.4006 30.5205
DWT 8.4049 9.3407 28.8218
AES + FADING CHANNEL
FFT 8.6932 14.1673 36.8018
DCT 8.5829 10.2435 14.1012
DWT 8.3737 9.2508 13.7209
RC6 + AWGN CHANNEL
FFT 8.3916 9.4165 30.9051
DCT 8.3961 9.3528 33.1261
DWT 8.4344 9.3419 29.1988
RC6 + FADING CHANNEL
FFT 8.7540 13.8893 40.0734
DCT 8.5691 10.2276 14.1490
DWT 8.3983 9.2497 13.9828
 Simul ati o n E x a m p l e s 19 5

90
Original image with CFO compensation
80 Chaotic encryption with CFO compensation
DES encryption with CFO compensation
AES encryption with CFO compensation
70
RC6 encryption with CFO compensation
Original without CFO compensation
60 Chaotic encryption without CFO compensation
DES encryption without CFO compensation
PSNR(dB)

50 AES encryption without CFO compensation

RC6 encryption without CFO compensation
40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)
(a) FFT-OFDM system.

90
Original image with CFO compensation
80 Chaotic encryption with CFO compensation
DES encryption with CFO compensation
AES encryption with CFO compensation
70
RC6 encryption with CFO compensation
Original without CFO compensation
60 Chaotic encryption without CFO compensation
DES encryption without CFO compensation
PSNR(dB)

50 AES encryption without CFO compensation

RC6 encryption without CFO compensation
40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)
(b) DCT-OFDM system.

Figure 9.91  Variation of the PSNR of the decrypted image with the Eb /N0 in AWGN channel: (a)
FFT-OFDM system; (b) DCT-OFDM system.
19 6 Im ag e En c ry p ti o n

90
Original image with CFO compensation
80 Chaotic encryption with CFO compensation
DES encryption with CFO compensation
AES encryption with CFO compensation
70
RC6 encryption with CFO compensation
Original without CFO compensation
60 Chaotic encryption without CFO compensation
DES encryption without CFO compensation
PSNR(dB)

50 AES encryption without CFO compensation

RC6 encryption without CFO compensation
40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.91 (Continued)  Variation of the PSNR of the decrypted image with the Eb /N0 in AWGN
channel: DWT-OFDM system.

9.3

9.2

9.1

9 Original image
Chaotic image encryption
8.9 DES image encryption
PSNR(dB)

AES image encryption

8.8 RC6 image encryption

8.7

8.6

8.5

8.4

8.3
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.92  PSNR versus Eb /N0 in the FFT-OFDM system for a Raleigh channel with fd =300 Hz
without CFO compensation.
 Simul ati o n E x a m p l e s 19 7

90

80 Original image
Chaotic image encryption
70 DES image encryption
AES image encryption
60 RC6 image encryption
PSNR(dB)

50

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.93  PSNR versus Eb /N0 in the FFT-OFDM system for a Raleigh channel with fd = 300 Hz
with CFO compensation.

9.3

9.2

9.1

8.9
PSNR(dB)

8.8

8.7
Original image
8.6
Chaotic image encryption
8.5 DES image encryption
AES image encryption
8.4 RC6 image encryption

8.3
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.94  PSNR versus Eb /N0 in the DCT-OFDM system for a Raleigh channel with fd = 300 Hz
without CFO compensation.
19 8 Im ag e En c ry p ti o n

90

80 Original image
Chaotic image encryption
70 DES image encryption
AES image encryption
60 RC6 image encryption
PSNR(dB)

50

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.95  PSNR versus Eb /N0 in the DCT-OFDM system for a Raleigh channel with fd = 300 Hz
with CFO compensation.

9.3

9.2

9.1

8.9
PSNR(dB)

8.8

8.7

8.6 Original image

Chaotic image encryption
8.5 DES image encryption
AES image encryption
8.4 RC6 image encryption

8.3
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.96  PSNR versus Eb /N0 in the DWT-OFDM system for a Raleigh channel with fd = 300 Hz
without CFO compensation.
 Simul ati o n E x a m p l e s 19 9

90

80 Original image
Chaotic image encryption
70 DES image encryption
AES image encryption
60 RC6 image encryption
PSNR(dB)

50

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.97  PSNR versus Eb /N0 in the DWT-OFDM system for a Raleigh channel with fd = 300 Hz
with CFO compensation.

80

Chaotic-ECB mode
70
Chaotic-CFB mode
Chaotic-CBC mode
60 Chaotic-OFB mode

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.98  PSNR versus Eb /N0 in the FFT-OFDM system for an AWGN channel with N = 128
and G = 32.
200 Im ag e En c ry p ti o n

80

70 Chaotic-ECB mode
Chaotic-CFB mode
Chaotic-CBC mode
60
Chaotic-OFB mode

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.99  PSNR versus Eb /N0 in the DCT-OFDM system for an AWGN channel with N = 128
and G = 32.

80
Chaotic-ECB mode
70 Chaotic-CFB mode
Chaotic-CBC mode
Chaotic-OFB mode
60

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.100  PSNR versus Eb /N0 in the DWT-OFDM system for an AWGN channel with N = 128
and G = 32.
 Simul ati o n E x a m p l e s 2 01

9.3

9.2

9.1

9
Chaotic-ECB mode
8.9 Chaotic-CFB mode
PSNR(dB)

Chaotic-CBC mode
8.8
Chaotic-OFB mode

8.7

8.6

8.5

8.4

8.3
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.101  PSNR versus Eb /N0 in the FFT-OFDM system for a Raleigh channel with fd = 300 Hz
without CFO compensation.

9.3

9.2

9.1

8.9 Chaotic-ECB mode

PSNR(dB)

Chaotic-CFB mode
8.8 Chaotic-CBC mode
Chaotic-OFB mode
8.7

8.6

8.5

8.4

8.3
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.102  PSNR versus Eb /N0 in the DCT-OFDM system for a Raleigh channel with fd = 300 Hz
without CFO compensation.
202 Im ag e En c ry p ti o n

9.3

9.2

9.1

9
Chaotic-ECB mode
8.9
Chaotic-CFB mode
PSNR(dB)

Chaotic-CBC mode
8.8
Chaotic-OFB mode

8.7

8.6

8.5

8.4

8.3
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.103  PSNR versus Eb /N0 in the DWT-OFDM system for a Raleigh channel with fd = 300 Hz
without CFO compensation.

80

Chaotic-ECB mode
70
Chaotic-CFB mode
Chaotic-CBC mode
60 Chaotic-OFB mode

50
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.104  PSNR versus Eb /N0 in the FFT-OFDM system for a Raleigh channel with fd = 300
Hz with CFO compensation.
 Simul ati o n E x a m p l e s 203

80

70
Chaotic-ECB mode
60 Chaotic-CFB mode
Chaotic-CBC mode
50 Chaotic-OFB mode
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.105  PSNR versus Eb /N0 in the DCT-OFDM system for a Raleigh channel with fd = 300
Hz with CFO compensation.

80

70

Chaotic-ECB mode
60 Chaotic-CFB mode
Chaotic-CBC mode
50 Chaotic-OFB mode
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.106  PSNR versus Eb /N0 in the DWT-OFDM system for a Raleigh channel with fd = 300 Hz
with CFO compensation.
204 Im ag e En c ry p ti o n

9.3

9.2

9.1

8.9
PSNR(dB)

8.8

8.7

8.6 Chaotic-HGI-ECB mode

Chaotic-HGI-CFB mode
Chaotic-HGI-CBC mode
8.5 Chaotic-HGI-OFB mode
Chaotic-ZP-ECB mode
Chaotic-ZP-CFB mode
8.4 Chaotic-ZP-CBC mode
Chaotic-ZP-OFB mode
8.3
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.107  PSNR versus Eb /N0 in the FFT-OFDM system with ZP and HGI for a Raleigh channel
without PSE and equalization at N = 128 and G = 32.

9.3

9.2

9.1

8.9
PSNR(dB)

8.8

8.7

8.6 Chaotic-HGI-ECB mode

Chaotic-HGI-CFB mode
Chaotic-HGI-CBC mode
8.5 Chaotic-HGI-OFB mode
Chaotic-ZP-ECB mode
8.4 Chaotic-ZP-CFB mode
Chaotic-ZP-CBC mode
Chaotic-ZP-OFB mode
8.3
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.108  PSNR versus Eb /N0 in the DCT-OFDM system with ZP and HGI for a Raleigh channel
without PSE and equalization at N = 128 and G = 32.
 Simul ati o n E x a m p l e s 205

9.3

9.2

9.1

8.9
PSNR(dB)

8.8

8.7

8.6 Chaotic-HGI-ECB mode

Chaotic-HGI-CFB mode
Chaotic-HGI-CBC mode
8.5 Chaotic-HGI-OFB mode
Chaotic-ZP-ECB mode
8.4 Chaotic-ZP-CFB mode
Chaotic-ZP-CBC mode
Chaotic-ZP-OFB mode
8.3
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.109  PSNR versus Eb /N0 in the DWT-OFDM system with ZP and HGI for a Raleigh chan-
nel without PSE and equalization at N = 128 and G = 32.

80
Chaotic-HGI-ECB mode
70 Chaotic-HGI-CFB mode
Chaotic-HGI-CBC mode
Chaotic-HGI-OFB mode
60
Chaotic-ZP-ECB mode
Chaotic-ZP-CFB mode
50 Chaotic-ZP-CBC mode
Chaotic-ZP-OFB mode
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.110  PSNR versus Eb /N0 in the FFT-OFDM system with ZP and HGI for a Raleigh channel
with PSE and equalization at N = 128 and G = 32.
206 Im ag e En c ry p ti o n

80
Chaotic-HGI-ECB mode
70 Chaotic-HGI-CFB mode
Chaotic-HGI-CBC mode
Chaotic-HGI-OFB mode
60
Chaotic-ZP-ECB mode
Chaotic-ZP-CFB mode
50 Chaotic-ZP-CBC mode
Chaotic-ZP-OFB mode
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.111  PSNR versus Eb /N0 in the DCT-OFDM system with ZP and HGI for a Raleigh channel
with PSE and equalization at N = 128 and G = 32.

80
Chaotic-HGI-ECB mode
70 Chaotic-HGI-CFB mode
Chaotic-HGI-CBC mode
Chaotic-HGI-OFB mode
60 Chaotic-ZP-ECB mode
Chaotic-ZP-CFB mode
50 Chaotic-ZP-CBC mode
Chaotic-ZP-OFB mode
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.112  PSNR versus Eb /N0 in the DWT-OFDM system with ZP and HGI for a Raleigh channel
with PSE and equalization at N = 128 and G = 32.
 Simul ati o n E x a m p l e s 207

80
Chaotic-ECB mode without CFO compensation
70 Chaotic-CFB mode without CFO compensation
Chaotic-CBC mode without CFO compensation
Chaotic-OFB mode without CFO compensation
60 Chaotic-ECB mode with CFO compensation
Chaotic-CFB mode with CFO compensation
Chaotic-CBC mode with CFO compensation
50 Chaotic-OFB mode with CFO compensation
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.113  PSNR versus Eb /N0 in the FFT-OFDM system for an AWGN channel.

80
Chaotic-ECB mode without CFO compensation
70 Chaotic-CFB mode without CFO compensation
Chaotic-CBC mode without CFO compensation
Chaotic-OFB mode without CFO compensation
60 Chaotic-ECB mode with CFO compensation
Chaotic-CFB mode with CFO compensation
Chaotic-CBC mode with CFO compensation
50 Chaotic-OFB mode with CFO compensation
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.114  PSNR versus Eb /N0 in the DCT-OFDM system for an AWGN channel.
208 Im ag e En c ry p ti o n

80
Chaotic-ECB mode without CFO compensation
Chaotic-CFB mode without CFO compensation
70 Chaotic-CBC mode without CFO compensation
Chaotic-OFB mode without CFO compensation
60 Chaotic-ECB mode with CFO compensation
Chaotic-CFB mode with CFO compensation
Chaotic-CBC mode with CFO compensation
50 Chaotic-OFB mode with CFO compensation
PSNR(dB)

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.115  PSNR versus Eb /N0 in the DWT-OFDM system for an AWGN channel.

50
AES + FFT – OFDM
45 AES with preprocessing + FFT – OFDM
AES + DCT – OFDM
40 AES with preprocessing + DCT – OFDM
AES + DWT – OFDM
35 AES with preprocessing + DWT – OFDM

30
PSNR(dB)

25

20

15

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.116  PSNR versus Eb /N0 in the OFDM system for an AWGN channel with G = 32 and
N = 128.
 Simul ati o n E x a m p l e s 209

10.2
AES + FFT – OFDM
10 AES with preprocessing + FFT – OFDM
AES + DCT – OFDM
AES with preprocessing + DCT – OFDM
9.8 AES + DWT – OFDM
AES with preprocessing + DWT – OFDM
9.6
PSNR(dB)

9.4

9.2

8.8

8.6

8.4
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.117  PSNR versus Eb /N0 in the OFDM system for a Raleigh channel without PSE and
equalization at N = 128 and G = 32.

50
AES+FFT-OFDM
45 AES with preprocessing+FFT-OFDM
AES+DCT-OFDM
40 AES with preprocessing+DCT-OFDM
AES+DWT-OFDM
35 AES with preprocessing+DWT-OFDM

30
PSNR(dB)

25

20

15

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.118  PSNR versus Eb /N0 in the OFDM system for a Raleigh channel with PSE and equal-
ization at N = 128 and G = 32.
210 Im ag e En c ry p ti o n

10.2 AES with preprocessing + FFT – OFDM

AES with preprocessing + DCT – OFDM
10 AES with preprocessing + DWT – OFDM
AES + FFT – OFDM
AES + DCT – OFDM
9.8 AES + DWT – OFDM

9.6
PSNR(dB)

9.4

9.2

8.8

8.6

8.4
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.119  PSNR versus Eb /N0 in the OFDM system for an AWGN channel without CFO
compensation.

50
AES with preprocessing + FFT – OFDM
45 AES with preprocessing + DCT – OFDM
AES with preprocessing + DWT – OFDM
40 AES + FFT – OFDM
AES + DCT – OFDM
AES + DWT – OFDM
35

30
PSNR(dB)

25

20

15

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.120  PSNR versus Eb /N0 in the OFDM system for an AWGN channel with CFO
compensation.
 Simul ati o n E x a m p l e s 211

50
AES with preprocessing + FFT – OFDM
45 AES with preprocessing + DCT – OFDM
AES with preprocessing + DWT – OFDM
40 AES + FFT – OFDM
AES + DCT – OFDM
35 AES + DWT – OFDM

30
PSNR(dB)

25

20

15

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.121  PSNR versus Eb /N0 in the OFDM system with HGI for a Raleigh channel with N =
128 and G = 32 with PSE and equalization.

50
AES with preprocessing + FFT – OFDM
45 AES with preprocessing + DCT – OFDM
AES with preprocessing + DWT – OFDM
40 AES + FFT – OFDM
AES + DCT – OFDM
35 AES + DWT – OFDM

30
PSNR(dB)

25

20

15

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.122  PSNR versus Eb /N0 in the DWT-OFDM system with ZP for a Raleigh channel with
N = 128 and G = 32 with PSE and equalization.
212 Im ag e En c ry p ti o n

50
RC6 with preprocessing + FFT – OFDM
45 RC6 with preprocessing + DCT – OFDM
RC6 with preprocessing + DWT – OFDM
40 RC6 + FFT – OFDM
RC6 + DCT – OFDM
35 RC6 + DWT – OFDM

30
PSNR(dB)

25

20

15

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.123  PSNR versus Eb /N0 in the OFDM system for an AWGN channel with N = 128 and
G = 32.

10.2 RC6 with preprocessing + FFT – OFDM

RC6 with preprocessing + DCT – OFDM
10 RC6 with preprocessing + DWT – OFDM
RC6 + FFT – OFDM
9.8 RC6 + DCT – OFDM
RC6 + DWT – OFDM
9.6
PSNR(dB)

9.4

9.2

8.8

8.6

8.4
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.124  PSNR versus Eb /N0 in the OFDM system for a Raleigh channel with fd = 300 Hz
without PSE and equalization.
 Simul ati o n E x a m p l e s 213

50
RC6 with preprocessing + FFT – OFDM
45 RC6 with preprocessing + DCT – OFDM
RC6 with preprocessing + DWT – OFDM
40 RC6 + FFT – OFDM
RC6 + DCT – OFDM
RC6 + DWT – OFDM
35

30
PSNR(dB)

25

20

15

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.125  PSNR versus Eb /N0 in the OFDM system for a Raleigh channel with fd = 300 Hz
with PSE and equalization.

10.2 RC6 with preprocessing + FFT – OFDM

RC6 with preprocessing + DCT – OFDM
10 RC6 with preprocessing + DWT – OFDM
RC6 + FFT – OFDM
9.8 RC6 + DCT – OFDM
RC6 + DWT – OFDM
9.6
PSNR(dB)

9.4

9.2

8.8

8.6

8.4
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.126  PSNR versus Eb /N0 in the OFDM system for an AWGN channel without CFO
compensation.
214 Im ag e En c ry p ti o n

50

45 RC6 with preprocessing + FFT – OFDM

RC6 with preprocessing + DCT – OFDM
RC6 with preprocessing + DWT – OFDM
40 RC6 + FFT – OFDM
RC6 + DCT – OFDM
35 RC6 + DWT – OFDM

30
PSNR(dB)

25

20

15

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.127  PSNR versus Eb /N0 in the OFDM system for an AWGN channel with CFO
compensation.

50
RC6 with preprocessing + FFT – OFDM
45 RC6 with preprocessing + DCT – OFDM
RC6 with preprocessing + DWT – OFDM
40 RC6 + FFT – OFDM
RC6 + DCT – OFDM
35 RC6 + DWT – OFDM

30
PSNR(dB)

25

20

15

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.128  PSNR versus Eb /N0 in the OFDM system for a Raleigh channel without CFO
compensation.
 Simul ati o n E x a m p l e s 215

50
RC6 with preprocessing + FFT – OFDM
45 RC6 with preprocessing + DCT – OFDM
RC6 with preprocessing + DWT – OFDM
40 RC6 + FFT – OFDM
RC6 + DCT – OFDM
35 RC6 + DWT – OFDM

30
PSNR(dB)

25

20

15

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.129  PSNR versus Eb /N0 in the OFDM system for a Raleigh channel with CFO
compensation.

100
Without coding
90 With clipping
With companding at µ = 0.1
80 With clipping + companding at µ = 0.1
With companding at µ = 4
70 With clipping + companding at µ = 4
PSNR(dB)

60

50

40

30

20

10
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.130  PSNR versus Eb /N0 for original-image transmission with FFT-OFDM system over an
AWGN channel. Convolutional coding is applied with all cases except the first one.
216 Im ag e En c ry p ti o n

100
Without coding
90 With clipping
With companding at µ = 0.1
80 With clipping + companding at µ = 0.1
With companding at µ = 4
70 With clipping + companding at µ = 4
PSNR(dB)

60

50

40

30

20

10
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.131  PSNR versus Eb /N0 for original-image transmission with DCT-OFDM system over
an AWGN channel. Convolutional coding is applied with all cases except the first one.

100
Without coding
90 With clipping
With companding at µ = 0.1
80 With clipping + companding at µ = 0.1
With companding at µ = 4
70 With clipping + companding at µ = 4

60
PSNR(dB)

50

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.132  PSNR versus Eb /N0 for original-image transmission with DWT-OFDM system over
an AWGN channel. Convolutional coding is applied with all cases except the first one.
 Simul ati o n E x a m p l e s 217

100
Without coding
90 With clipping
With companding at µ = 0.1
80 With clipping + companding at µ = 0.1
With companding at µ = 4
70 With clipping + companding at µ = 4
PSNR(dB)

60

50

40

30

20

10
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.133  PSNR versus Eb /N0 for chaotic-encrypted image transmission with FFT-OFDM
system over an AWGN channel. Convolutional coding is applied with all cases except the first one.

100
Without coding
90 With clipping
With companding at µ = 0.1
80 With clipping + companding at µ = 0.1
With companding at µ = 4
70 With clipping + companding at µ = 4
PSNR(dB)

60

50

40

30

20

10
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.134  PSNR versus Eb /N0 for chaotic-encrypted image transmission with DCT-OFDM
system over an AWGN channel. Convolutional coding is applied with all cases except the first one.
218 Im ag e En c ry p ti o n

90
Without coding
80 With clipping
With companding at µ = 0.1
70 With clipping + companding at µ = 0.1
With companding at µ = 4
60 With clipping + companding at µ = 4
PSNR(dB)

50

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.135  PSNR versus Eb /N0 for chaotic-encrypted image transmission with DWT-OFDM
system over an AWGN channel. Convolutional coding is applied with all cases except the first one.

100
Without coding
90 With clipping
With companding at µ = 0.1
80 With clipping + companding at µ = 0.1
With companding at µ = 4
70 With clipping + companding at µ = 4

60
PSNR(dB)

50

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.136  PSNR versus Eb /N0 for AES-encrypted image transmission with FFT-OFDM system
over an AWGN channel. Convolutional coding is applied with all cases except the first one.
 Simul ati o n E x a m p l e s 219

100
Without coding
90 With clipping
With companding at µ = 0.1
80 With clipping + companding at µ= 0.1
With companding at µ =4
70 With clipping + companding at µ = 4

60
PSNR(dB)

50

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.137  PSNR versus Eb /N0 for AES-encrypted image transmission with DCT-OFDM system
over an AWGN channel. Convolutional coding is applied with all cases except the first one.

100
Without coding
90
With clipping
With companding at µ = 0.1
80 With clipping+companding at µ = 0.1
With companding at µ = 4
70 With clipping+companding at µ = 4
60
PSNR(dB)

50

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.138  PSNR versus Eb /N0 for AES-encrypted image transmission with DWT-OFDM system
over an AWGN channel. Convolutional coding is applied with all cases except the first one.
220 Im ag e En c ry p ti o n

100
Without coding
90 With clipping
With companding at µ = 0.1
80 With clipping + companding at µ = 0.1
With companding at µ = 4
70 With clipping + companding at µ = 4

60
PSNR(dB)

50

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.139  PSNR versus Eb /N0 for DES-encrypted image transmission with FFT-OFDM system
over an AWGN channel. Convolutional coding is applied with all cases except the first one.

100

90 Without coding
With clipping
80 With companding at µ = 0.1
With clipping + companding at µ = 0.1
70 With companding at µ = 4
With clipping + companding at µ = 4
60
PSNR(dB)

50

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.140  PSNR versus Eb /N0 for DES-encrypted image transmission with DCT-OFDM system
over an AWGN channel. Convolutional coding is applied with all cases except the first one.
 Simul ati o n E x a m p l e s 2 21

100
Without coding
90
With clipping
With companding at µ = 0.1
80
With clipping + companding at µ = 0.1
With companding at µ= 4
70
With clipping + companding at µ = 4

60
PSNR(dB)

50

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.141  PSNR versus Eb /N0 for DES-encrypted image transmission with DWT-OFDM system
over an AWGN channel. Convolutional coding is applied with all cases except the first one.

100

90 Without coding
With clipping
80 With companding at µ = 0.1
With clipping + companding at µ = 0.1
70 With companding at µ = 4
With clipping + companding at µ = 4
60
PSNR(dB)

50

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.142  PSNR versus Eb /N0 for RC6-encrypted image transmission with FFT-OFDM system
over an AWGN channel. Convolutional coding is applied with all cases except the first one.
222 Im ag e En c ry p ti o n

100
Without coding
90 With clipping
With companding at µ = 0.1
80 With clipping + companding at µ = 0.1
With companding at µ = 4
70 With clipping + companding at µ = 4

60
PSNR(dB)

50

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.143  PSNR versus Eb /N0 for RC6-encrypted image transmission with DCT-OFDM system
over an AWGN channel. Convolutional coding is applied with all cases except the first one.

100
Without coding
90
With clipping
With companding at µ = 0.1
80
With clipping + companding at µ= 0.1
With companding at µ = 4
70 With clipping + companding at µ = 4
60
PSNR(dB)

50

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Eb/N0(dB)

Figure 9.144  PSNR versus Eb /N0 for RC6-encrypted image transmission with DWT-OFDM sys-
tem over an AWGN channel. Convolutional coding is applied with all cases except the first one.
 Simul ati o n E x a m p l e s 223

9.5 Summary

From all obtained results, we can see that PAPR reduction technique
effects can be mitigated to a great extent with error-correcting codes
such as the convolutional code. Another conclusion is that chaotic
encryption preserves its rank as the best encryption algorithm from
the immunity to noise perspective even in the presence of PAPR
reduction. A simple clipping, companding, or hybrid clipping-and-
companding method may be feasible for PAPR reduction in the case
of encrypted image transmission.
References

1. O. S. Faragallah, Utilization of Security Techniques for Multimedia

Applications, PhD thesis, Department of Computer Science and
Engineering, Faculty of Electronic Engineering, Menoufiya University,
Egypt, 2007.
2. A. J. Menezes, P. C. V. Oorschot, and S. Vanstone, Handbook of Applied
Cryptography, CRC Press, Boca Raton, FL, 1996.
3. L. Qiao, Multimedia Security and Copyright Protection, PhD thesis,
Department of Computer Science, University of Illinois at Urbana-
Champaign, 1998.
4. W. Stallings, Cryptography and Network Security Principles and Practice,
Prentice-Hall, Upper Saddle River, NJ, 1999.
5. C. E. Shannon, Communication Theory of Secrecy Systems, Bell System
Technical Journal, Vol. 28, No. 4, pp. 656–715, October 1949.
6. S. Li, G. Chen, and X. Zheng, Chaos-Based Encryption for Digital
Images and Videos, Chapter 4 in Multimedia Security Handbook, CRC
Press, Boca Raton, FL, February 2004.
7. Y. Mao and M. Wu, A Joint Signal Processing and Cryptographic
Approach to Multimedia Encryption, IEEE Transactions on Image
Processing, Vol. 15, No. 7, pp. 2061–2075, July 2006.
8. Y. Mao,Research on Chaos-Based Image Encryption and Watermarking
Technology, PhD thesis, Department of Automation, Nanjing University
of Science and Technology, China, August 2003.
9. J. Daemen and V. Rijmen, AES Proposal: Rijndael, AES Algorithm
Submission to the National Institute of Standards and Technology, 1999.
10. R. L. Rivest, M. J. B. Robshaw, R. Sidney, and Y. L. Yin, The RC6TM Block
Cipher, MIT Laboratory for Computer Science, Cambridge, MA, 1998.
11. R. F. Sewell, Bulk Encryption Algorithm for Use with RSA, Electronics
Letters, Vol. 29, No. 25, pp. 2183–2185, December 9, 1993.

225
226 Ref eren c e s

12. H. E. H. Ahmed, H. M. Kalash, and O. S. Faragallah, Encryption

Efficiency Analysis and Security Evaluation of RC6 Block Cipher
for Digital Images, International Conference on Electrical Engineering
(ICEE ‘07), pp. 1–7, 11–12, April 2007.
13. C. E. Shannon, A Mathematical Theory of Communication, Bell System
Technical Journal, Vol. 27, pp. 379–423, July 1948.
14. M. Asim and V. Jeoti, On Image Encryption: Comparison between
AES and a Novel Chaotic Encryption Scheme, International Conference
on Signal Processing, Communications and Networking (ICSCN ‘07),
pp. 65–69, 22–24, February 2007.
15. R. Matthews, On the Derivation of a Chaotic Encryption Algorithm,
Cryptologia, Vol. 8, No. 1, pp. 29–41, 1989.
16. B. Schneier, Applied Cryptography, 2nd edition, Wiley, New York, 1996.
17. L. Kocarev, Chaos-Based Cryptography: A Brief Overview, IEEE Circuits
and Systems Magazine, Vol. 1, No. 3, pp. 6–21, 2001.
18. D. Stinson, Cryptography: Theory and Practice, 2nd edition, Chapman &
Hall/CRC Press, Boca Raton, FL, 2002.
19. Y. Mao, G. Chen, and S. Lian, A Novel Fast Image Encryption Scheme
Based on 3D Chaotic Baker Maps, International Journal of Bifurcation
and Chaos, Vol. 14, No. 10, pp. 3613–3624, 2004.
20. S. Li, Analyses and New Designs of Digital Chaotic Ciphers, PhD t­ hesis,
School of Electronics and Information Engineering, Xi’an Jiaotong
University, Xi’an, China, June 2003.
21. National Bureau of Standards, Data Encryption Standard Modes of
Operation, Federal Information Processing Standards Publication 81,
U.S. Government Printing Office, Washington, DC, 1980.
22. S. Li, X. Mou, and Y. Cai, Pseudo-Random Bit Generator Based on
Couple Chaotic Systems and Its Applications in Stream Cipher
Cryptography, Lecture Notes in Computer Science, Vol. 2247, pp.
316–329, 2001.
23. B. A. Forouzan, Cryptography and Network Security, McGraw-Hill, New
Delhi, India, 2007.
24. R. Kusters and M. Tuengerthal, Universally Composable Symmetric
Encryption, 2nd IEEE Computer Security Foundations Symposium (CSF ‹09),
pp. 293–307, July 2009.
25. H. Jin, Z. Liao, D. Zou, and C. Li, Asymmetrical Encryption Based
Automated Trust Negotiation Model, 2nd IEEE International Conference
on Digital Ecosystems and Technologies (DEST 2008), pp. 363–368,
February 2008.
26. S. G. Lian, J. Sun, and Z. Wang, A Novel Image Encryption Scheme
Based on JPEG Encoding, Proceedings of 8th International Conference on
Information Visualization, pp. 217–220, July 14–16, 2004.
27. M. V. Droogenbroeck and R. Benedett, Techniques for a Selective
Encryption of Uncompressed and Compressed Images, Proceedings of
Advanced Concepts for Intelligent Vision Systems (ACIVS), Ghent, Belgium,
pp. 90–97, September 9–11, 2002.
 Ref eren c e s 227

28. F. Dachselt, K. Kelber, and W. Schwarz, Chaotic Coding and

Cryptoanalysis, Proceedings of IEEE International Symposium on Circuits
and Systems, Hong Kong, pp. 1061–1064, June 9–12, 1997.
29. S. Li and X. Zheng, Cryptanalysis of a Chaotic Image Encryption
Method, Proceedings of IEEE International Symposium on Circuits and
Systems (ISCAS), Vol. 2, pp. 708–711, 2002.
30. J. Wei, X. Liao, K. W. Wong, and T. Zhou, Cryptanalysis of Cryptosystem
Using Multiple One-Dimensional Chaotic Maps, Communications in
Nonlinear Science and Numerical Simulation, Vol. 12, pp. 814–822, 2007.
31. H. Gao, Y. Zhang, S. Liang, and D. Li, A New Chaotic Algorithm for
Image Encryption, Chaos, Solitons & Fractals, Vol. 29, No. 2, pp. 393–399,
July 2006.
32. X. Li, J. Knipe, and H. Cheng, Image Compression and Encryption
Using Tree Structures, Pattern Recognition Letters, Vol. 18, No. 8,
pp. 2439–2451,1997.
33. J. I. Guo, J. C. Yen, and J. C. Yeh, The Design and Realization of a
New Hierarchical Chaotic Image Encryption Algorithm, Proceedings of
the International Symposium on Communications (ISCOM’99), pp. 210–
214, 1999.
34. L. Zhang, X. Liao, and X. Wang, An Image Encryption Approach Based
on Chaotic Maps, Chaos, Solitons & Fractals, Vol. 24, No. 3, pp. 759–765,
May 2005.
35. L. Kocarev and G. Jakimoski, Logistic Map as a Block Encryption
Algorithm, Physics Letters A, Vol. 289, No. 4–5, pp. 199–206, October 22,
2001.
36. T. Xiang, X. Liao, G. Tang, Y. Chen, and K.W. Wong, A Novel Block
Cryptosystem Based on Iterating a Chaotic Map, Physics Letters A,
Vol. 349, No. 1–4, pp. 109–115, January 9, 2006.
37. S. Contini, R. L. Rivest, M. J. B. Robshaw, and Y. L. Yin, The Security of
the RC6TM Block Cipher, Version 1.0, RSA Laboratories, MIT Laboratory
for Computer Science, Cambridge, MA, August 20, 1998.
38. M. Salleh, S. Ibrahim, and I. F. Isnin, Enhanced Chaotic Image
Encryption Algorithm Based on Baker›s Map, Proceedings of the 2003
International Symposium on Circuits and Systems (ISCAS ‹03), Vol. 2,
pp. 508–511, May 2003.
39. D. Chen, A Feasible Chaotic Encryption Scheme for Image, International
Workshop on Chaos-Fractals Theories and Applications (IWCFTA›09),
pp. 172–176, November 6–85, 2009.
40. A. Palacios and H. Juarez, Cryptography with Cycling Chaos, Physics
Letters A, Vol. 303, No. 5–6, pp. 345–351, October 28, 2002.
41. H. Cheng and X. Li, Partial Encryption of Compressed Images and
Videos, IEEE Transactions on Signal Processing, Vol. 48, No. 8, pp. 2439–
2451, August 2000.
42. A. Servetti and J. C. De Martin, Perception-Based Partial Encryption of
Compressed Speech, IEEE Transactions on Speech and Audio Processing,
Vol. 10, No. 8, pp. 637–643, November 2002.
228 Ref eren c e s

43. X. Wu, H. Hu, and B. Zhang, Analyzing and Improving a Chaotic

Encryption Method, Chaos, Solitons & Fractals, Vol. 22, No. 2, pp.
367–373, October 2004.
44. R. C. Merkle and M. Hellman, On the Security of Multiple Encryption,
Communications of the ACM, Vol. 24, No. 7, pp. 465–467, July 1981.
45. P. Kitsos, S. Goudevenos, and O. Koufopavlou, VLSI Implementations of
the Triple-DES Block Cipher, Proceedings of the 10th IEEE International
Conference on Electronics, Circuits and Systems (ICECS), Vol. 1, pp. 76–79,
December 14–17, 2003.
46. National Bureau of Standards, Data Encryption Standard, Federal
Information Processing Standards Publication 46, U.S. Government
Printing Office, Washington, DC, 1977.
47. W. Jian, L. Xu, and J. Xiaoyong, A Secure Communication System with
Multiple Encryption Algorithms, Proceedings of the International Conference
on E-Business and E-Government, pp. 3574–3577, May 7–9, 2010.
48. R. L. Rivest, The RC5 Encryption Algorithm, Proceedings of the 2nd
Workshop on Fast Software Encryption, pages 86–96, 1995.
49. R. Baldwin and R. Rivest, RFC 2040: The RC5, RC5-CBC, RC5-CBC-
Pad, and RC5-CTS Algorithms, October 30, 1996.
50. R. Ratchkov, Advanced Encryption Standard (AES) Algorithm and Modes of
Operation, LSI Logic, San Jose, CA, August 2002.
51. J. C. Yen and J. I. Guo, A New Hierarchical Chaotic Image Encryption
Algorithm and Its Hardware Architecture, Proceedings 9th VLSI Design/
CAD Symposium, Taiwan, pp. 358–362, 1998.
52. J. Fridrich, Symmetric Ciphers Based on Two-Dimensional Chaotic
Maps, International Journal of Bifurcation and Chaos, Vol. 8, No. 6,
pp. 1259–1284, 1998.
53. W. Xiao, J. Zhang, and W. Wu, A Watermarking Algorithm Based
on Chaotic Encryption, Proceedings of IEEE TENCON, pp. 545–548,
2002.
54. S. Li and X. Zheng, On the Security of an Image Encryption Method,
Proceedings IEEE International Conference on Image Processing (ICIP),
Vol. 2, pp. 925–928, 2002.
55. M. Ashtiyani, P. M. Birgani, and H. M. Hosseini, Chaos-Based Medical
Image Encryption Using Symmetric Cryptography, 3rd International
Conference on Information and Communication Technologies: From Theory to
Applications (ICTTA), pp. 1–5, 7–11 April 2008.
56. N. Bourbakis and C. Alexopoulos, Picture Data Encryption Using SCAN
Patterns, Pattern Recognition, Vol. 25, No. 6, pp. 567–581, 1992.
57. J. Scharinger, Fast Encryption of Image Data Using Chaotic Kolmogrov
Flow, Journal of Electronic Engeering, Vol. 7, No. 2, pp. 318–325, 1998.
58. J. C. Yen and J. I. Guo, A New Image Encryption Algorithm and Its
VLSI Architecture, Proceedings of the IEEE Workshop Signal Processing
Systems, pp. 430– 437, 1999.
59. J. C. Yen and J. I. Guo, A New Chaotic Key Based Design for Image
Encryption and Decryption, Proceedings of the IEEE International
Symposium Circuits and Systems, Vol. 4, pp. 49–52, 2000.
 Ref eren c e s 229

60. G. Chen, Y. Mao, and C. K. Chui, A Symmetric Image Encryption

Scheme Based on 3D Chaotic Cat Maps, Chaos, Solitons and Fractals,
Vol. 21, No. 3, pp. 749–761, 2004.
61. K. Kelber and W. Schwarz, General Design Rules for Chaos-Based
Encryption Systems, International Symposium on Nonlinear Theory and
Its Applications (NOLTA2005), pp. 465–468, Bruges, Belgium, October
18–21, 2005.
62. M. Yang, N. Bourbakis, and S. Li, Data-Image-Video Encryption,
Potentials IEEE, Vol. 23, No. 3, pp. 28–34, 2004.
63. Y. Feng and X. Yu, A Novel Symmetric Image Encryption Approach
Based on an Invertible Two-Dimensional Map, 35th Annual Conference
of IEEE Industrial Electronics (IECON ‘09), pp. 1973–1978, November
3–5, 2009.
64. C. J. Kuo, Novel Image Encryption Technique and Its Application in
Progressive Transmission, Journal of Electronic Imaging, Vol. 2, No. 4,
pp. 345–351, 1993.
65. H. K. Chang and J. L. Liou, An Image Encryption Scheme Based on
Quadtree Compression Scheme, Proceedings of the International Computer
Symposium, Taiwan, pp. 230–237, 1994.
66. J. Scharinger, Secure and Fast Encryption Using Chaotic Kolmogorov
Flows, Proceedings of the IEEE Industrial Electronics and Applications
Conference, pp. 3662–3666, 1998.
67. H. M. Elkamchouchi and M. A. Makar, Measuring Encryption Quality
for Bitmap Images Encrypted with Rijndael and KAMKAR Block
Ciphers, Proceedings of the National Radio Science Conference of Egypt,
pp. 277–284, 2005.
68. S.A. Yeung, S. Zhu, and B. Zeng, Quality Assessment for a Perceptual Video
Encryption System, Proceedings of the IEEE Wireless Communications,
Networking and Information Security (WCNIS) Conference, pp. 102–106,
2010.
69. A. Sinha and K. Singh, A Technique for Image Encryption Using Digital
Signature, Optics Commununication, Vol. 218, No. 4–6, pp. 229–234, 2003.
70. D. E. Newton, Encyclopedia of Cryptology, ABC-CLIO, Santa Barbara,
CA, 1997.
71. Z. Yun-Peng, L. Wei, C. Shui-Ping, Z. Zheng-Jun, N. Xuan, and D.
Wei-Di, Digital Image Encryption Algorithm Based on Chaos and
Improved DES, IEEE International Conference on Systems, Man and
Cybernetics, pp. 474–479, 2009.
72. J. Cheng, F. Zhang, K. Yu, and J. Ma, The Dynamic and Double Encryption
System Based on Two-Dimensional Image, International Conference on
Computational Intelligence and Security (CIS ’09), pp. 458–462, 2009.
73. A. H. M. Ragab, N. A. Ismail, and O. S. FaragAllah, Enhancements and
Implementation of RC6 Block Cipher for Data Security, Proceedings
of International Conference on Electrical and Electronic Technology, Vol. 1,
pp. 133–137, 2001.
74. H. Feistel, Cryptography and Computer Privacy, Scientific American,
Vol. 228, No. 5, pp. 15–23, May 1973.
230 Ref eren c e s

75. H. H. Nien, S. K. Changchien, S. Y. Wu, and C. K. Huang, A New

Pixel-Chaotic-Shuffle Method for Image Encryption, 10th International
Conference on Control, Automation, Robotics and Vision (ICARCV), pp.
883–887, December 17–20, 2008.
76. H. M. Heys Analysis of the Statistical Cipher Feedback Mode of Block
Ciphers, IEEE Transactions on Computers, Vol. 52, No. 1, pp. 77–92,
January 2003.
77. C. Lu and S. Tseng, Integrated Design of AES (Advanced Encryption
Standard) Encrypter and Decrypter, Proceedings of The IEEE International
Conference on Application-Specific Systems, Architectures and Processors, pp.
277–285, July 2002.
78. M. Borsc and H. Shinde, Wireless Security & Privacy, IEEE International
Conference on Personal Wireless Communications (ICPWC), pp. 424–428,
January 2005.
79. A. G. Chefranov and T. A. Mazurova, Pseudo-Random Number
Generator (RC4) Period Improvement, International Conference
on  Automation, Quality and Testing, Robotics, Vol. 2, pp. 38–41,
May 2006.
80. I. F. Elashry, O. S. Farag Allah, A. M. Abbas, S. El-Rabaie, and F. E. Abd
El-Samie, Homomorphic Image Encryption, Electronic Imaging, Vol. 18,
No. 3, 033002, 2009.
81. W. K. Wong, L. P. Lee, and K. W. Wong, A Modified Chaotic
Cryptographic Method, Computer Physics Communication, Vol. 13, No. 8,
pp. 234–236, 2001.
82. M. Ahmad, C. Gupta, and A. Varshney, Digital Image Encryption
Based  on Chaotic Map for Secure Transmission, International
Multimedia, Signal Processing and Communication Technologies (IMPACT
’09), pp. 292–295, March 2009.
83. D. Garg and S. Verma, Improvement Over Public Key Cryptographic
Algorithm, IEEE International Advance Computing Conference (IACC),
pp. 734–739, March 2009.
84. A. Ramzi, A. N. El-Kassar, and B. M. Shebaro, A Comparative Study of
Elgamal Based Digital Signature Algorithms, World Automation Congress
(WAC ’06), pp. 1–6, July 2006.
85. G. Kim, J. Kim, and G. Cho, An Improved RC6 Algorithm with the Same
Structure of Encryption and Decryption, 11th International Conference
on Advanced Communication Technology (ICACT), Vol. 2, pp. 1211–1215,
February 2009.
86. Q. Zhu, L.Li, J. Liu, and N. Xu, The Analysis and Design of Accounting
Information Security System Based on AES Algorithm, International
Conference on Machine Learning and Cybernetics, Vol. 5, pp. 2713–2718,
July 2009.
87. S.C. Koduru and V. Chandrasekaran, Integrated Confusion-Diffusion
Mechanisms for Chaos-Based Image Encryption, IEEE 8th International
Conference on Computer and Information Technology Workshops, pp. 260–
263, July 2008.
 Ref eren c e s 2 31

88. S. A. N. Gilani and M. A. Bangash, Enhanced Block Based Color Image

Encryption Technique with Confusion, IEEE International Multitopic
Conference (INMIC), pp. 200–206, December 2008.
89. D. Coppersmith, The Data Encryption Standard (DES) and Its Strength
Against Attacks, IBM Journal of Research and Development, Vol. 38, No. 3,
pp. 243–250, 1994.
90. J. Daemen and V. Rijmen, Advanced Encryption Standard (AES), FIPS
197, Technical report, Catholic University, ESAT, Leuven, Belgium,
November 2001.
91. W. Meier and L. R. Knudsen, Correlations in RC6 with a Reduced
Number of Rounds Source, Proceedings of the 7th International Workshop
on Fast Software Encryption, pp. 94–108, 2000.
92. K. L. Chung and L. C. Chang, Large Encrypting Binary Images with
Higher Security, Pattern Recognition Letters, Vol. 19, No. 5–6, pp. 461–
468, April 1998.
93. A. M. Fiskiran and R. B. Lee, Performance Impact of Addressing Modes
on Encryption Algorithms, Proceedings of the International Conference on
Computer Design (ICCD), pp. 542–545, September 2001.
94. M. Fouad, D. H. Salem, and I. Ziedan, Application of Data Encryption
Standard to Bitmap and JPEG Images, Proceedings of the Twentieth
National Radio Science Conference (NRSC), pp. 1–8, March 2003.
95. W. Ying, Z. W. Zhao, and Z. Lelin, A Fault-Tolerable Encryption
Algorithm for Two-Dimensional Digital Image, 2nd IEEE Conference
on Industrial Electronics and Applications (ICIEA), pp. 2737–2741, May
23–25, 2007.
96. S. J. Shyu, Image Encryption by Random Grids, Pattern Recognition, Vol.
40, No. 3, pp. 1014–1031, 2007.
97. H. El-Din, H. Ahmed, H. M. Kalash, and O. S. Faragallah, An Efficient
Chaos Based Feedback Stream Cipher (ECBFSC) for Image Encryption
and Decryption, Informatica, Vol. 31, No.1, pp. 121–129, 2007.
98. S. Li, X. Zheng, X. Mou, and Y. Cai, Chaotic Encryption Scheme for
Real-Time Digital Video, Proceedings of SPIE, Vol. 4666, pp. 149–160,
2002.
99. S. Li, X. Mou, and Y. Cai, Improving Security of a Chaotic Encryption
Approach, Physics Letters A, Vol. 290, No. 3–4, pp. 127–133, November
12, 2001.
100. F. Han, X. Yu and S. Han, Improved Baker Map for Image Encryption,
1st International Symposium on Systems and Control in Aerospace and
Astronautics (ISSCAA), pp. 1273–1276, January 19–21, 2006.
101. S. Lian, J. Sun, and Z. Wang, Security Analysis of a Chaos-Based Image
Encryption Algorithm, Physica A: Statistical and Theoretical Physics,
Vol. 351, No. 2–4, pp. 645–661, June 15, 2005.
102. N. K. Pareek, V. Patidar, and K. K. Sud, Cryptography Using Multiple
One-Dimensional Chaotic Maps, Communications in Nonlinear
Science and Numerical Simulation, Vol. 10, No. 7, pp. 715–723, October
2005.
232 Ref eren c e s

103. I. F. Elashry, O. S. Faragallah, A. M. Abbas, S. El-Rabaie, and F. E. Abd

El-Samie, A New Method for Encrypting Images with Few Details
Using Rijndael and RC6 Block Ciphers in the Electronic Code Book
Mode, Information Security Journal: A Global Perspective, Vol. 21, No, 4, pp.
193–205, 2012.
104. C. A. Henk and V. Tilborg, Encyclopedia of Cryptography and Security,
Springer Science and Business Media, New York, 2005.
105. N. El-Fishawy and O. M. Abu Zaid, Quality of Encryption Measurement
of Bitmap Images with RC6, MRC6, and Rijndael Block Cipher
Algorithms, International Journal of Network Security, Vol. 5, No. 3,
pp. 241–251, 2007.
106. X. Wang and D. Zhao, Image Encryption Based on Anamorphic Fractional
Fourier Transform and Three-Step Phase-Shifting Interferometry, Optics
Communications, Vol. 268, No. 2, pp. 240–244, December 15, 2006.
107. S. Behnia, A. Akhshani, H. Mahmodi, and A. Akhavan, A Novel
Algorithm for Image Encryption Based on Mixture of Chaotic Maps,
Chaos, Solitons & Fractals, Vol. 35, No. 2, pp. 408–419, January 2008.
108. Y. B. Mao and G. Chen, Chaos-Based Image Encryption, in Handbook
of Computational Geometry for Pattern Recognition, Computer Vision,
Neuralcomputing and Robotics, Springer-Verlag, Berlin, pp. 231–265,
2005.
109. C. C. Chang, M. S. Hwang, and T. S. Chen, A New Encryption Algorithm
for Image Cryptosystems, Journal of Systems and Software, Vol. 58, No. 2,
pp. 83–91, 2001.
110. D. C. Key and J. R. Levine, Graphics File Format, Winderest Books/
McGraw-Hill, New York, 1992.
111. D. Kim and G. L. Stüber, Residual ISI Cancellation for OFDM with
Applications to HDTV Broadcasting, IEEE Journal on Selected Areas in
Communications, Vol. 16, No. 8, 1590–1599, October 1998.
112. G. H. Yang, D. Shen, and V. O. K. Li, UEP for Video Transmission in
Space-Time Coded OFDM Systems, IEEE Infocom, 2004.
113. M. I. Rahman, S. S. Das, and Frank H.P. Fitzek, OFDM Based WLAN
Systems, Technical Report R-04-1002, Center for TeleInFrastruktur
(CTiF), Copenhagen, Denmark, Vol. 1.2, February 18, 2005.
114. E. P. Lawrey, Adaptive Techniques for Multiuser OFDM, PhD thesis,
James Cook University, Townsville, QLD, Australia, December 2001.
115. E. Lawrey, The Suitability of OFDM as a Modulation Technique
for  Wireless Telecommunications, with a CDMA Comparison,
Bachelor thesis, James Cook University, Townsville, QLD, Australia,
October 1997.
116. K. Abdullah and Z. M. Hussain, Performance of Fourier-Based and
Wavelet-Based OFDM for DVB-T Systems, Proceeding of the 2007
Australasian Telecommunication Networks and Applications Conference,
Christchurch, New Zealand, December 2nd–5th 2007.
117. J. J. van de Beek, O. Edfors, M. Sandell, S. K. Wilson, and P. O. Borjesson,
On Channel Estimation in OFDM Systems, IEEE 45th Vehicular
Technology Conference, 1995.
 Ref eren c e s 233

118. P. Tan and N. C. Beaulieu, A Comparison of DCT-Based OFDM and

DFT-Based OFDM in Frequency Offset and Fading Channels, IEEE
Transactions on Communications, Vol. 54, No. 11, 2113–2125, November
2006.
119. J. G. Andrews, A. Ghosh, and R. Muhamed, Fundamentals of
WiMAX Understanding Broadband Wireless Networking, Prentice Hall
Communications Engineering and Emerging Technologies Series,
Prentice-Hall, Englewood Cliffs, NJ, pp. 113–145, February 2007.
120. A. K. Lee, Ooi, M. Drieberg, and V. Jeoti, DWT based FFT in Practical
OFDM Systems, IEEE 2006.
121. P. Liu, B. B. Li, Z. Y. Lu, and F. K. Gong, An OFDM Bandwidth
Estimation Scheme for Spectrum Monitoring, IEEE 2005.
122. X. F. Wang, Y. R. Shayan, and M. Zeng, On the Code and Interleaver
Design of Broadband OFDM Systems, IEEE Communications Letters,
Vol. 8, No. 11, 653–655, November 2004.
123. F. Gao, T. Cui, A. Nallanathan, and C. Tellambura, Maximum Likelihood
Based Estimation of Frequency and Phase Offset in DCT OFDM
Systems under Non Circular Transmissions: Algorithms, Analysis and
Comparisons, IEEE Transactions on Communications, Vol. 56, No. 9,
1425–1429, September 2008.
124. R. Merched, On OFDM and Signal-Carrier Frequency-Domain Systems
Based on Trigonometric Transforms, IEEE Signal Processing Letters, Vol.
13, No. 8, 473–476, August 2006.
125. E. Lawrey and C. J. Kikkert, Peak to Average Power Ratio Reduction
of OFDM Signals Using Peak Reduction Carriers, Fifth International
Symposium on Signal Processing and its Applications, ISSPA ’99, Brisbane,
Australia, August 22–25, 1999.
126. M. Shen, G. Li, and H. Liu, Effect of Traffic Channel Configuration
on the Orthogonal Frequency Division Multiple Access Downlink
Performance, IEEE Transactions on Wireless Communications, Vol. 4,
No. 4, 1901–1913, July 2005.
127. H. Schulze and C. Luders, Theory and Application of OFDM and CDMA
Wideband Wireless Communications, Wiley New York, pp. 145–264, 2005.
128. N. Al-Dhahir and H. Minn, A New Multicarrier Transceiver Based
on the Discrete Cosine Transform, Proceedings of the IEEE Wireless
Communications and Networking Conference, Vol. 1, pp. 45–50, March
13–17, 2005.
129. P. Tan and N. C. Beaulieu, An Improved DCT-Based OFDM Data
Transmission Scheme, Proceedings of the IEEE 16th PIMRC’05, 2005.
130. H. Harada and R. Prasad, Simulation and Software Radio for Mobile
Communications, House Universal Personal Communications Library, 2002.
131. M. S. El-Tanany, Y. Wu, and L. Házy, OFDM Uplink for Interactive
Broadband Wireless: Analysis and Simulation in the Presence of Carrier,
Clock and Timing Errors, IEEE, 2001.
132. A. Langowski, Time and Frequency Synchronisation in 4G OFDM
Systems, EURASIP Journal on Wireless Communications and Networking,
doi:10.1155/2009/641292, 2009.
234 Ref eren c e s

133. Mary Ann Ingram, OFDM Simulation Using Matlab, Smart Antenna
Research Laboratory, Guillermo Acosta, Georgia Tech, Atlanta,
August 2000.
134. J. Zhang and B. Li, New Modulation Identification Scheme for OFDM
in Multipath Rayleigh Fading Channel, International Symposium on
Computer Science and Computational Technology, 2008.
135. H. S. Chu, B. S. Park, C. K. An, J. S. Kang, and H. G. Son, Wireless Image
Transmission based on Adaptive OFDM System, IEEE, 2007.
136. C. S. Avilaf and R. S. Reillot, The Rijndael Block Cipher (AES Proposal):
A Comparison with DES, IEEE, 2001.
137. M. S. Liu, Y. Zhang, and J. Huali, Research on Improving Security of DES
by Chaotic Mapping, Proceedings of the Eighth International Conference on
Machine Learning and Cybernetics, Baoding, July 12–15.
138. V. B. Vats, K. K. Garg, and A. Abad, Performance Analysis of DFT-
OFDM, DCT-OFDM, and DWT-OFDM Systems in AWGN,
Proceedings of the IEEE Fourth International Conference on Wireless and
Mobile Communications, 2008.
139. P. Tan and N. C. Beaulieu, Precise Bit Error Probability Analysis of
DCT OFDM in the Presence of Carrier Frequency Offset on AWGN
Channels, Proceedings of the IEEE Globcom 2005, pp. 1429–1434, 2005.
140. Giridhar D. Mandyam, On the Discrete Cosine Transform and OFDM
Systems, Nokia Research Center, Irving, TX, 2003.
141. M. Misiti, Y. Misiti, and G. Oppenheim, Wavelets and Their Applications,
Jean-Michel Poggi, ISTE, 2007.
142. A. Prochazka, J. Uhlir, P. J. W. Rayner, and N. J. Kingsbury, Signal Analysis
and Prediction, Birkhauser, New York, 1998.
143. B. Muquet, Z. Wang, G. B. Giannakis, M. d. Courville, and P.
Duhamel, Cyclic Prefixing or Zero Padding for Wireless Multicarrier
Transmissions, IEEE Transactions on Communications, Vol. 50, No. 12,
December 2002.
144. B. Li, S. Zhou, M. Stojanovic, L. Freitag, and P. Willett, Non-Uniform
Doppler Compensation for Zero-Padded OFDM over Fast-Varying
Under water Acoustic Channels, OCEANS2007-Europe, IEEE, 2007.
145. C. R. N. Athaudage and R. R. V. Angiras, Sensitivity of FFT-Equalised
Zero-Padded OFDM Systems to Time and Frequency Synchronization
Errors, IEE Proc.-Commun., Vol. 152, No. 6, December 2005.
146. B. Muquet, M. D. Courville, P. Duhamel, G. B. Giannakis, and
P.  Magniez, Turbo Demodulation of Zero-Padded OFDM
Transmissions, IEEE Transactions on Communications, Vol. 50, No. 11,
November 2002.
147. D. Huang and K. B. Letaief, An Interference-Cancellation Scheme
for Carrier Frequency Offsets Correction in OFDMA Systems, IEEE
Transactions on Communication, Vol. 53, No. 7, pp. 1155–1165, July 2005.
148. J. Paul and M. G. Linnartz, Performance Analysis of Synchronous
MC-CDMA in Mobile Rayleigh Channel with Both Delay and
Doppler Spreads, IEEE Transactions on Vehicular Technology, Vol. 50,
No. 6, November 2001.
 Ref eren c e s 235

149. T. Cui, F. Gao, A. Nallanathan, and C. Tellambura, ML CFO and PO

Estimation in DCT OFDM Systems under Non-Circular Transmissions,
IEEE Communications Society, 2007.
150. P. Bansal and A. Brzezinski, Adaptive Loading in MIMO/OFDM
Systems, December 13, 2001.
151. H. Hwang and H. Park, Doppler Frequency Offset Estimation in OFDM
Systems, Mobile Telecommunication Research Division, ETRI, IEEE,
2009.
152. J. Lee, D. Toumpakaris, H. L. Lou, and J. M. Cioffi, Effect of Carrier
Frequency Offset on Time-Domain Differential Demodulation in
OFDM Systems, IEEE, 2004.
153. M. Anandpara, E. Erwa, J. Golab, R. Samanta, and H. Wang, Inter-carrier
Interference Cancellation for OFDM Systems, EE 381K-11: Wireless
Communications, May 6, 2003.
154. A. S. Baiha, M. Singh, A. J. Goldsmith, and B. R. Saltzberg, A New
Approach for Evaluating Clipping Distortion in Multicarrier Systems,
IEEE Journal on Select Areas in Communications, Vol. 20, No. 5, pp. 1037–
1046, June 2002.
155. R. V. Nee and R. Prasad, OFDM for Wireless Multimedia Communications,
Artech House, Boston, 2000.
156. R. V. Nee and A, D. Wild, Reducing the Peak-to-Average Power Ratio of
OFDM, Proceedings of the IEEE Vehicular Technology Conference (VTC’98),
pp. 2072–2076, May 1998.
157. S. Wei, D. L. Goeckel, and P. E. Kelly, A Modern Extreme Value Theory
Scheme to Calculating the Distribution of the PAPR in OFDM
Systems, Proceedings of IEEE ICC 2002, New York, pp. 1686–90,
May 2002.
158. M. Sharif, M. Gharavi-Alkhansari, and B. H. Khalaj, On the Peak-
to-Average Power of OFDM Signals Based on Oversampling, IEEE
Transactions on Communications, Vol. 51, pp. 72–78, January 2003.
159. G. Wunder and H. Boche, Upper Bounds on the Statistical Distribution
of the Crestfactor in OFDM Transmission, IEEE Transactions on
Information Theory, Vol. 49, pp. 488–494, February 2003.
160. R. Prasad, OFDM for Wireless Communications Systems, Artech House,
Boston, 2004.
161. S. H. Han and J. H. Lee, Modified Selected Mapping Scheme for PAPR
Reduction of Coded OFDM Signal, IEEE Transactions on Broadcasting,
Vol. 50, pp. 335–341, September 2004.
162. H. Dai and H. V. Poor, Advanced Signal Processing for Power Line
Communications, IEEE Communications Magazine, Vol. 41, No. 5,
pp. 100–107, May 2003.
163. S. B. Weinstein, The History of Orthogonal Frequency Division
Multiplexing, IEEE Communications Magazine, pp. 26–35, November
2009.
164. S. Celebi, Interblock Interference (IBI) Minimizing Time-Domain
Equalizer (TEQ) for OFDM, IEEE Signal Processing Letters, Vol. 10,
No. 8, pp. 232–234, August 2003.
236 Ref eren c e s

165. Y. Wu and W. Y. Zou, Orthogonal Frequency Division Multiplexing:

A Multicarrier Modulation Scheme, IEEE Transactions on Consumer
Electronics, Vol. 41, No. 3, pp. 392–399, August 1995.
166. T. Pollet, P. Spruyt, and M. Moeneclaey, The BER Performance of
OFDM Systems Using Non-Synchronized Sampling, Proceedings of
GLOBECOM, 1994, pp. 253–257.
167. ETSI, Transmission and Multiplexing (TM); Access Transmission Systems
on Metallic Access Cables; Very High Speed Digital Subscriber Line (VDSL);
Part 2: Transceiver Specification, 2002, TS 101 270-2.
168. F. J. Cañete, J. A. Cortées, L. Díez, and J. T. Entrambasaguas, Modeling
and Evaluation of the Indoor Power Line Transmission Medium, IEEE
Communications Magazine, Vol. 41, No. 4, pp. 41–47, April 2003.
Appendix A

function y = multi(x,w)
%z = a*a1 mod 2^32
a = frombase256(x); a1 = frombase256(w);
z = mod(a*a1,2^32);
y = tobase256(z);

function permutation = p(substitution)

% p function permutes an input "substitution" based on x
x = [16 7 20 21 29 12 28 17 01 15 23 26 05 18 31 10 2
8 24 14 32 27 3 9 19 13 30 6 22 11 4 25];
permutation = substitution(x);

function y = padding(x)
%Padding x with zeros if x is shorter than 16
L = length(x);
if mod(L,16)==0
y = x;
else
pad = 16-mod(L,16);
y = zeros(1,L+pad);
y(1:L) = x;
end;
function RC5enctextRC5 = RC5CBC(plaintext1,m,n,CO,r,​
keyi)
% Encrypt data using RC5 in CBC mode.
RC5enctextRC5 = zeros(m*n/8,8);

237
238 App en d i x A

for i = 1:m*n/8
plaintext = plaintext1(i,:);
if i==1
plaintexts = bitxor(plaintext,CO);
else
plaintexts = bitxor(plaintext,RC5enctext);
end;
RC5enctext = RC5enc(plaintexts,r,keyi);
RC5enctextRC5(i,:) = RC5enctext;

end;

function plaintextRC5 = RC5CBCDec(RC5enctext1,m,n,CO,r,​

keyi)
% Decrypt data using RC5 in CBC mode.
plaintextRC5 = zeros(m*n/8,8);
for i = 1:m*n/8
RC5enctextss = RC5enctext1(i,:);
re_plaintext = RC5decry(RC5enctextss,r,keyi);
if i==1
plaintext = bitxor(re_plaintext,CO);
else
 plaintext = bitxor(re_plaintext,RC5enctext1
((i-1),:));
end;
plaintextRC5(i,:) = plaintext;
end;

function RC5enctextRC5 = RC5CFB(plaintext1,m,n,CO,r,​

keyi)
% Encrypt data using RC5 in CFB mode.
RC5enctextRC5 = zeros(m*n/8,8);
for i = 1:m*n/8
plaintext = plaintext1(i,:);
if i==1
cr1 = RC5enc(CO,r,keyi);

RC5enctext = bitxor(plaintext,cr1);
else
cr1 = RC5enc(RC5enctext,r,keyi);

RC5enctext = bitxor(plaintext,cr1);
end;
RC5enctextRC5(i,:) = RC5enctext;

end;
 App en d i x A 239

function plaintextRC5 = RC5CFBDec(RC5enctext1,m,n,CO,r,​

keyi)
% Decrypt data using RC5 in CFB mode.
plaintextRC5 = zeros(m*n/8,8);
for i = 1:m*n/8
RC5enctextss = RC5enctext1(i,:);
if i==1
cr1 = RC5enc(CO,r,keyi);

re_plaintext = bitxor(RC5enctextss,cr1);
else
cr1 = RC5enc(RC5enctext1(i-1,:),r,keyi);

re_plaintext = bitxor(RC5enctextss,cr1);
end;

plaintextRC5(i,:) = re_plaintext;
end;

function [xa1, Fs, nbits] = RC5DecCBCAudio(x1, Fs,

nbits,key,r,CO)
% Decrypt Audio using RC5 in CBC mode.
keyi = RC5keygen(key,r);
%CBC L = length(x1); y1 = double(x1);
plaintextRC5 = zeros(L/8,8);
RC5enctext1 = reshape(y1,L/8,8);
for i = 1:L/8
RC5enctextss = RC5enctext1(i,:);
re_plaintext = RC5decry(RC5enctextss,r,keyi);
if i==1

plaintext = bitxor(re_plaintext,CO);
else
 plaintext = bitxor(re_plaintext,RC5enctext1
((i-1),:));
end;
plaintextRC5(i,:) = plaintext;

end;
xa1 = uint8(plaintextRC5(:));

function xa1 = RC5DecCBCImage(x1,key,r,CO)

% Decrypt Image using RC5 in CBC mode.
keyi = RC5keygen(key,r);
%CBC
24 0 App en d i x A

[m,n] = size(x1);
y1 = double(x1);
plaintextRC5 = zeros(m*n/8,8);
RC5enctext1 = reshape(y1',8,m*n/8)';
for i = 1:m*n/8
RC5enctextss = RC5enctext1(i,:);
re_plaintext = RC5decry(RC5enctextss,r,keyi);
if i = =1

plaintext = bitxor(re_plaintext,CO);
else
 plaintext = bitxor(re_plaintext,RC5enctext1
((i-1),:));
end;
plaintextRC5(i,:) = plaintext;

end;
ya1 = reshape(plaintextRC5',n,m)';

xa1 = uint8(ya1);

function xa1 = RC5DecCBCimageC(x,key,r,CO)

% Decrypt Colored Image using RC5 in CBC mode.
keyi = RC5keygen(key,r);
%CBC
x1 = x(:,:,1);
x2 = x(:,:,2);
x3 = x(:,:,3);
y1 = double(x1);
y2 = double(x2);
y3 = double(x3);
[m,n] = size(x1);
plaintext(:,:,1) = reshape(y1',8,m*n/8)';
plaintext(:,:,2) = reshape(y2',8,m*n/8)';
plaintext(:,:,3) = reshape(y3',8,m*n/8)';
ciphertext = x;
for i = 1:3 plaintextRC5 = plaintext(:,:,i);
 RC5enctext = RC5CBCDec(plaintextRC5,m,n,CO,
r,keyi);
RC5enctext = reshape(RC5enctext',n,m)';
ciphertext(:,:,i) = RC5enctext;
end

xa1 = uint8(ciphertext);
 App en d i x A 2 41

function [xa1, Fs, nbits] = RC5DecCFBAudio(x1, Fs,

nbits,key,r,CO)
%Decrypt Audio using RC5 in CFB mode.
keyi = RC5keygen(key,r);
L = length(x1); y1 = double(x1);
plaintextRC5 = zeros(L/8,8);
RC5enctext1 = reshape(y1,L/8,8);
for i = 1:L/8
RC5enctextss = RC5enctext1(i,:);
if i = =1
cr1 = RC5enc(CO,r,keyi);

re_plaintext = bitxor(RC5enctextss,cr1);
else
cr1 = RC5enc(RC5enctext1(i-1,:),r,keyi);

re_plaintext = bitxor(RC5enctextss,cr1);
end;

plaintextRC5(i,:) = re_plaintext;
end;
xa1 = uint8(plaintextRC5(:));

function xa1 = RC5DecCFBImage(x1,key,r,CO)

%Decrypt Image using RC5 in CFB mode.
keyi = RC5keygen(key,r);
[m,n] = size(x1); y1 = double(x1);
plaintextRC5 = zeros(m*n/8,8);
RC5enctext1 = reshape(y1',8,m*n/8)';
for i = 1:m*n/8
RC5enctextss = RC5enctext1(i,:);
if i = =1
cr1 = RC5enc(CO,r,keyi);

re_plaintext = bitxor(RC5enctextss,cr1);
else
cr1 = RC5enc(RC5enctext1(i-1,:),r,keyi);
re_plaintext = bitxor(RC5enctextss,cr1);
end;

plaintextRC5(i,:) = re_plaintext;
end;
ya1 = reshape(plaintextRC5',n,m)';

xa1 = uint8(ya1);
242 App en d i x A

function xa1 = RC5DecCFBImageC(x,key,r,CO)

%Decrypt Colored Image using RC5 in CFB mode.
keyi = RC5keygen(key,r);
%CFB

x1 = x(:,:,1);
x2 = x(:,:,2);
x3 = x(:,:,3);
y1 = double(x1);
y2 = double(x2);
y3 = double(x3);
[m,n] = size(x1);
plaintext(:,:,1) = reshape(y1',8,m*n/8)';
plaintext(:,:,2) = reshape(y2',8,m*n/8)';
plaintext(:,:,3) = reshape(y3',8,m*n/8)';
ciphertext = x;
for i = 1:3 plaintextRC5 = plaintext(:,:,i);
 RC5enctext = RC5CFBDec(plaintextRC5,m,n,CO,r,​
keyi);
RC5enctext = reshape(RC5enctext',n,m)';
ciphertext(:,:,i) = RC5enctext;
end

xa1 = uint8(ciphertext);

function [xa1, Fs, nbits] = RC5DecECBAudio(x1, Fs,

nbits,key,r)
%Decrypt Audio using RC5 in ECB mode.
keyi = RC5keygen(key,r);
%ECB L = length(x1);
y1 = double(x1);
plaintextRC5 = zeros(L/8,8);
RC5enctext1 = reshape(y1,L/8,8);
for i = 1:L/8
RC5enctextss = RC5enctext1(i,:);
re_plaintext = RC5decry(RC5enctextss,r,keyi);
plaintextRC5(i,:) = re_plaintext;

end;
xa1 = uint8(plaintextRC5(:));

function xa1 = RC5DecECBImage(x1,key,r)

%Decrypt Image using RC5 in ECB mode.
keyi = RC5keygen(key,r);
 App en d i x A 24 3

[m,n] = size(x1); y1 = double(x1);

plaintextRC5 = zeros(m*n/8,8);
RC5enctext1 = reshape(y1',8,m*n/8)';
for i = 1:m*n/8
RC5enctextss = RC5enctext1(i,:);
re_plaintext = RC5decry(RC5enctextss,r,keyi);
plaintextRC5(i,:) = re_plaintext;

end;
ya1 = reshape(plaintextRC5',n,m)';

xa1 = uint8(ya1);

function xa1 = RC5DecECBImageC(x,key,r)

%Decrypt Colored Image using RC5 in ECB mode.
keyi = RC5keygen(key,r);
%ECB
x1 = x(:,:,1);
x2 = x(:,:,2);
x3 = x(:,:,3);
y1 = double(x1);
y2 = double(x2);
y3 = double(x3);
[m,n] = size(x1);
plaintext(:,:,1) = reshape(y1',8,m*n/8)';
plaintext(:,:,2) = reshape(y2',8,m*n/8)';
plaintext(:,:,3) = reshape(y3',8,m*n/8)';
ciphertext = x;
for i = 1:3 plaintextRC5 = plaintext(:,:,i);
RC5enctext = RC5ECBDec(plaintextRC5,m,n,r,keyi);
RC5enctext = reshape(RC5enctext',n,m)';
ciphertext(:,:,i) = RC5enctext;
end
xa1 = uint8(ciphertext);

function [xa1, Fs, nbits] = RC5DecOFBAudio(x1, Fs,

nbits,key,r,CO)
%Decrypt Audio using RC5 in OFB mode.
keyi = RC5keygen(key,r);
%OFB

L = length(x1);
y1 = double(x1);
plaintextRC5 = zeros(L/8,8);
RC5enctext1 = reshape(y1,L/8,8);
24 4 App en d i x A

for i = 1:L/8
RC5enctextss = RC5enctext1(i,:);
if i = =1
cr = RC5enc (CO,r,keyi);
re_plaintext = bitxor(RC5enctextss,cr);
else cr = RC5enc(cr,r,keyi);
re_plaintext = bitxor(RC5enctextss,cr);
end;

plaintextRC5(i,:) = re_plaintext;

end;
xa1 = uint8(plaintextRC5(:));

function xa1 = RC5DecOFBImage(x1,key,r,CO)

%Decrypt Image using RC5 in OFB mode.
keyi = RC5keygen(key,r);

[m,n] = size(x1); y1 = double(x1);

plaintextRC5 = zeros(m*n/8,8);
RC5enctext1 = reshape(y1',8,m*n/8)';
for i = 1:m*n/8
RC5enctextss = RC5enctext1(i,:);
if i = =1
cr = RC5enc (CO,r,keyi);
re_plaintext = bitxor(RC5enctextss,cr);
else cr = RC5enc(cr,r,keyi);
re_plaintext = bitxor(RC5enctextss,cr);
end;

plaintextRC5(i,:) = re_plaintext;

end;
ya1 = reshape(plaintextRC5',n,m)';

xa1 = uint8(ya1);

function xa1 = RC5DecOFBImagec(x,key,r,CO)

%Decrypt Colored Image using RC5 in OFB mode.
keyi = RC5keygen(key,r);
%OFB

x1 = x(:,:,1);
x2 = x(:,:,2);
x3 = x(:,:,3);
 App en d i x A 24 5

y1 = double(x1);
y2 = double(x2);
y3 = double(x3);
[m,n] = size(x1);
plaintext(:,:,1) = reshape(y1',8,m*n/8)';
plaintext(:,:,2) = reshape(y2',8,m*n/8)';
plaintext(:,:,3) = reshape(y3',8,m*n/8)';
ciphertext = x;
for i = 1:3 plaintextRC5 = plaintext(:,:,i);
 RC5enctext = RC5OFBDec(plaintextRC5,m,n,CO,r,k
eyi);
RC5enctext = reshape(RC5enctext',n,m)';
ciphertext(:,:,i) = RC5enctext;
end

xa1 = uint8(ciphertext);

function y = RC5decry(plaintext,round,s)
%RC5 Decryption
a = plaintext(1:4); b = plaintext(5:8); for
i = round:-1:1
 b = RC5sub(b,s(2*i+2,:)); b = shifting​
(b,-LSB5(a)); b = bitxor(b',a);
 a = RC5sub(a,s(2*i+1,:)); a = shifting​
(a,-LSB5(b)); a = bitxor(a',b);
end b = RC5sub(b,s(2,:));
a = RC5sub(a,s(1,:)); y(1:4) = a;
y(5:8) = b;

function RC5enctextRC5 = RC5ECB(plaintext1,m,n,r,keyi)

%Encrypt data using RC5 in ECB mode
RC5enctextRC5 = zeros(m*n/8,8);
for i = 1:m*n/8 plaintext = plaintext1(i,:);
RC5enctext = RC5enc(plaintext,r,keyi);
RC5enctextRC5(i,:) = RC5enctext;

end;

function plaintextRC5 = RC5ECBDec(RC5enctext1,m,n,r,​

keyi)
%Decrypt data using RC5 in ECB mode
plaintextRC5 = zeros(m*n/8,8);
for i = 1:m*n/8
RC5enctextss = RC5enctext1(i,:);
re_plaintext = RC5decry(RC5enctextss,r,keyi);
24 6 App en d i x A

plaintextRC5(i,:) = re_plaintext;

end;

function y = RC5enc(plaintext,round,s)
%RC5 Encryption
a = plaintext(1:4); b = plaintext(5:8);
a = add(a,s(1,:)); b = add(b,s(2,:)); for
i = 1:round
a = bitxor(a,b); a = shifting(a,LSB5(b));
a = add(a,s(2*i+1,:)); b = bitxor(b,a);
b = shifting(b,LSB5(a)); b = add(b,s(2*i+2,:));
end y(1:4) = a; y(5:8) = b;

function [xa, Fs, nbits] = RC5EncCBCAudio(x, Fs,

nbits,key,r,CO)
%Encrypt Audio using RC5 in CBC mode
L = length(x);
y = double(x); keyi = RC5keygen(key,r);
plaintext1 = reshape(y,L/8,8);
RC5enctextRC5 = zeros(L/8,8);
%CBC
for i = 1:L/8 plaintext = plaintext1(i,:); if i = =1
plaintexts = bitxor(plaintext,CO);
else plaintexts = bitxor(plaintext,RC5enctext);
end;
RC5enctext = RC5enc(plaintexts,r,keyi);
RC5enctextRC5(i,:) = RC5enctext;

end;
xa = uint8(RC5enctextRC5(:));

function xa = RC5EncCBCImage(x,key,r,CO)
%Encrypt Image using RC5 in CBC mode
[m,n] = size(x);
y = double(x); keyi = RC5keygen(key,r);
plaintext1 = reshape(y',8,m*n/8)';

RC5enctextRC5 = zeros(m*n/8,8);
%CBC
for i = 1:m*n/8 plaintext = plaintext1(i,:); if i = =1
plaintexts = bitxor(plaintext,CO);
else plaintexts = bitxor(plaintext,RC5enctext);
end;
RC5enctext = RC5enc(plaintexts,r,keyi);
 App en d i x A 2 47

RC5enctextRC5(i,:) = RC5enctext;

nd; ya = reshape(RC5enctextRC5',n,m)';
e
xa =  uint8(ya);

function xa = RC5EncCBCImageC(x,key,r,CO)
%Encrypt Colored Image using RC5 in CBC mode

x1 = x(:,:,1); x2 = x(:,:,2); x3 = x(:,:,3);

y1 = double(x1); y2 = double(x2);
y3 = double(x3); [m,n] = size(x1);

keyi = RC5keygen(key,r);
plaintext(:,:,1) = reshape(y1',8,m*n/8)';
plaintext(:,:,2) = reshape(y2',8,m*n/8)';
plaintext(:,:,3) = reshape(y3',8,m*n/8)';

RC5enctext = x;
%CBC

for i = 1:3 plaintext1 = plaintext(:,:,i);

 RC5enctextRC5 = RC5CBC(plaintext1,m,n,CO,r,keyi);
RC5enctextRC5 = reshape(RC5enctextRC5',n,m)';
RC5enctext(:,:,i) = RC5enctextRC5;
end;
xa = uint8(RC5enctext);

function [xa, Fs, nbits] = RC5EncCFBAudio(x, Fs,

nbits,key,r,CO)
%Encrypt Audio using RC5 in CBC mode
L = length(x);
y = double(x); keyi = RC5keygen(key,r);
plaintext1 = reshape(y,L/8,8);
RC5enctextRC5 = zeros(L/8,8);
% CFB
for i = 1:L/8 plaintext = plaintext1(i,:); if i = =1
cr1 = RC5enc(CO,r,keyi);

RC5enctext = bitxor(plaintext,cr1);
else cr1 = RC5enc(RC5enctext,r,keyi);

RC5enctext = bitxor(plaintext,cr1);

end;
24 8 App en d i x A

RC5enctextRC5(i,:) = RC5enctext;

end;
xa = uint8(RC5enctextRC5(:));

function xa = RC5EncCFBImage(x,key,r,CO)
%Encrypt Image using RC5 in CFB mode
[m,n] = size(x);
y = double(x); keyi = RC5keygen(key,r);
plaintext1 = reshape(y',8,m*n/8)';

RC5enctextRC5 = zeros(m*n/8,8);
for i = 1:m*n/8 plaintext = plaintext1(i,:); if i = =1
cr1 = RC5enc(CO,r,keyi);

RC5enctext = bitxor(plaintext,cr1);
else cr1 = RC5enc(RC5enctext,r,keyi);

RC5enctext = bitxor(plaintext,cr1);
end;

RC5enctextRC5(i,:) = RC5enctext;

end; ya = reshape(RC5enctextRC5',n,m)';
xa = uint8(ya);

function xa = RC5EncCFBImageC(x,key,r,CO)
%Encrypt Colored Image using RC5 in CFB mode
x1 = x(:,:,1); x2 = x(:,:,2); x3 = x(:,:,3);
y1 = double(x1); y2 = double(x2);
y3 = double(x3); [m,n] = size(x1);

keyi = RC5keygen(key,r);
plaintext(:,:,1) = reshape(y1',8,m*n/8)';
plaintext(:,:,2) = reshape(y2',8,m*n/8)';
plaintext(:,:,3) = reshape(y3',8,m*n/8)';

RC5enctext = x;
% CFB
for i = 1:3 plaintext1 = plaintext(:,:,i);
RC5enctextRC5 = RC5CFB(plaintext1,m,n,CO,r,keyi);
RC5enctextRC5 = reshape(RC5enctextRC5',n,m)';
RC5enctext(:,:,i) = RC5enctextRC5;
end; xa = uint8(RC5enctext);
 App en d i x A 24 9

imwrite(xa,'onion1RC5encCFB.tif','tif')
imshow('onion1RC5encCFB.tif')

function [xa, Fs, nbits] = RC5EncECBAudio(x, Fs,

nbits,key,r)
%Encrypt Audio using RC5 in ECB mode
L = length(x);
y = double(x); keyi = RC5keygen(key,r);
plaintext1 = reshape(y,L/8,8);
RC5enctextRC5 = zeros(L/8,8);
% ECB
for i = 1:L/8 plaintext = plaintext1(i,:);
RC5enctext = RC5enc(plaintext,r,keyi);
RC5enctextRC5(i,:) = RC5enctext;

end;
xa = uint8(RC5enctextRC5(:));

function xa = RC5EncECBImage(x,key,r)
%Encrypt Image using RC5 in ECB mode
[m,n] = size(x);
y = double(x); keyi = RC5keygen(key,r);
plaintext1 = reshape(y',8,m*n/8)';

RC5enctextRC5 = zeros(m*n/8,8);
% ECB
for i = 1:m*n/8
plaintext = plaintext1(i,:);
RC5enctext = RC5enc(plaintext,r,keyi);
RC5enctextRC5(i,:) = RC5enctext;

nd; ya = reshape(RC5enctextRC5',n,m)';
e
xa = uint8(ya);

function xa = RC5EncECBImageC(x,key,r)
%Encrypt Colored Image using RC5 in ECB mode
x1 = x(:,:,1); x2 = x(:,:,2); x3 = x(:,:,3);
y1 = double(x1); y2 = double(x2);
y3 = double(x3); [m,n] = size(x1);

keyi = RC5keygen(key,r);

plaintext(:,:,1) = reshape(y1',8,m*n/8)';
plaintext(:,:,2) = reshape(y2',8,m*n/8)';
plaintext(:,:,3) = reshape(y3',8,m*n/8)';
250 App en d i x A

RC5enctext = x;% ECB

for i = 1:3 plaintext1 = plaintext(:,:,i);
RC5enctextRC5 = RC5ECB(plaintext1,m,n,r,keyi);
RC5enctextRC5 = reshape(RC5enctextRC5',n,m)';
RC5enctext(:,:,i) = RC5enctextRC5;
end;
xa = uint8(RC5enctext);

function [xa, Fs, nbits] = RC5EncOFBAudio(x, Fs,

nbits,key,r,CO)
%Encrypt Audio using RC5 in OFB mode
L = length(x);
y = double(x); keyi = RC5keygen(key,r);
plaintext1 = reshape(y,L/8,8);
RC5enctextRC5 = zeros(L/8,8);
% OFB
for i = 1:L/8 plaintext = plaintext1(i,:);

if i = =1 cr = RC5enc(CO,r,keyi);

else cr = RC5enc(cr,r,keyi);

end; RC5enctext = bitxor(plaintext,cr);

RC5enctextRC5(i,:) = RC5enctext;

end;
xa = uint8(RC5enctextRC5(:));

function xa = RC5EncOFBImage(x,key,r,CO)
%Encrypt Image using RC5 in OFB mode
[m,n] = size(x);
y = double(x); keyi = RC5keygen(key,r);
plaintext1 = reshape(y',8,m*n/8)';

RC5enctextRC5 = zeros(m*n/8,8);
% OFB
for i = 1:m*n/8 plaintext = plaintext1(i,:);

if i = =1 cr = RC5enc(CO,r,keyi);

else cr = RC5enc(cr,r,keyi);

end; RC5enctext = bitxor(plaintext,cr);

RC5enctextRC5(i,:) = RC5enctext;
 App en d i x A 2 51

end; ya = reshape(RC5enctextRC5',n,m)';
xa =  uint8(ya);

function xa = RC5EncOFBImageC(x,key,r,CO)
%Encrypt Colored Image using RC5 in OFB mode

x1 = x(:,:,1); x2 = x(:,:,2); x3 = x(:,:,3);

y1 = double(x1); y2 = double(x2);
y3 = double(x3); [m,n] = size(x1);
keyi = RC5keygen(key,r);
plaintext(:,:,1) = reshape(y1',8,m*n/8)';
plaintext(:,:,2) = reshape(y2',8,m*n/8)';
plaintext(:,:,3) = reshape(y3',8,m*n/8)';
RC5enctext = x;
% OFB
for i = 1:3 plaintext1 = plaintext(:,:,i);
 RC5enctextRC5 = RC5OFB(plaintext1,m,n,CO,r,keyi);
RC5enctextRC5 = reshape(RC5enctextRC5',n,m)';
RC5enctext(:,:,i) = RC5enctextRC5;
end;
xa = uint8(RC5enctext);

function s = keygen(key,r)
% RC5/RC6 Key Generation
p = [99 81 225 183];
q = [185 121 55 158];
s(1,:) = p;
for i = 2:2*r+4 s(i,:) = add(s(i-1,:),q);
end; i = 1; a = zeros(1,4);
b = zeros(1,4); v = 3*(2*r+4); for
h = 1:v
 s(i,:) = add(s(i,:),a); s(i,:) = add(s(i,:),b);
s(i,:) = shifting(s(i,:),3); a = s(i,:); key =
add(key,a); key = add(key,b);
key = shifting(key,LSB5(add(a,b)));
b = key;
if i = =2*r+4 i = 1;
end end

function RC5enctextRC5 = RC5OFB(plaintext1,m,n,CO,r,k

eyi)
%Encrypt Data using RC5 in OFB mode
RC5enctextRC5 = zeros(m*n/8,8);
for i = 1:m*n/8 plaintext = plaintext1(i,:);
252 App en d i x A

if i = =1 cr = RC5enc(CO,r,keyi);

else cr = RC5enc(cr,r,keyi);

end; RC5enctext = bitxor(plaintext,cr);

RC5enctextRC5(i,:) = RC5enctext;

end;

function plaintextRC5 = RC5OFBDec(RC5enctext1,m,n,CO,r,​

keyi)
%Decrypt data using RC5 in OFB mode

plaintextRC5 = zeros(m*n/8,8);
for i = 1:m*n/8
RC5enctextss = RC5enctext1(i,:);
if i = =1
cr = RC5enc (CO,r,keyi);
re_plaintext = bitxor(RC5enctextss,cr);
else cr = RC5enc(cr,r,keyi);
re_plaintext = bitxor(RC5enctextss,cr);
end;

plaintextRC5(i,:) = re_plaintext;

end;

function y = RC5sub(x,w)
%z = a-a1 mod 2^32
a = frombase256(x); a1 = frombase256(w);
z = mod(a-a1,2^32); y = tobase256(z);

function RC6enctextRC6 = CBC(plaintext1,m,n,CO,r,keyi)

%Encrypt Data using RC6 in CBC mode
RC6enctextRC6 = zeros(m*n/16,16);
for i = 1:m*n/16 plaintext = plaintext1(i,:); if i = =1
plaintexts = bitxor(plaintext,CO);
else plaintexts = bitxor(plaintext,RC6enctext);
end;
RC6enctext = RC6enc(plaintexts,r,keyi);
RC6enctextRC6(i,:) = RC6enctext;

end;
 App en d i x A 253

function plaintextRC6 = CBCDec(RC6enctext1,m,n,CO,

r,keyi)
%Decrypt data using RC6 in CBC mode

plaintextRC6 = zeros(m*n/16,16);
for i = 1:m*n/16
RC6enctextss = RC6enctext1(i,:);
re_plaintext = RC6dec(RC6enctextss,r,keyi);
if i = =1 plaintext = bitxor(re_plaintext,CO);
else
 plaintext = bitxor(re_plaintext,R
C6enctext1((i-1),:));
end;
plaintextRC6(i,:) = plaintext;
end;

function RC6enctextRC6 = CFB(plaintext1,m,n,CO,r,keyi)

%Encrypt data using RC6 in CFB mode

RC6enctextRC6 = zeros(m*n/16,16);
for i = 1:m*n/16 plaintext = plaintext1(i,:); if i = =1
cr1 = RC6enc(CO,r,keyi);

RC6enctext = bitxor(plaintext,cr1);
else cr1 = RC6enc(RC6enctext,r,keyi);

 RC6enctext = bitxor(plaintext,cr1); end;

RC6enctextRC6(i,:) = RC6enctext;

end;

function plaintextRC6 = CFBDec(RC6enctext1,m,n,CO,r,​

keyi)
%Decrypt data using RC6 in CFB mode
plaintextRC6 = zeros(m*n/16,16);
for i = 1:m*n/16
RC6enctextss = RC6enctext1(i,:);
if i = =1 cr1 = RC6enc(CO,r,keyi);

re_plaintext = bitxor(RC6enctextss,cr1);
else
cr1 = RC6enc(RC6enctext1(i-1,:),r,keyi);

re_plaintext = bitxor(RC6enctextss,cr1);
end;
25 4 App en d i x A

plaintextRC6(i,:) = re_plaintext;

end;

function y = RC6dec(plaintext,round,s)
% RC6 Decryption
a = plaintext(1:4); b = plaintext(5:8);
c = plaintext(9:12); d = plaintext(13:16);
c = RC6sub(c,s(2*round+4,:));
a = RC6sub(a,s(2*round+3,:)); for i = round:-1:1
temp = d; d = c; c = b; b = a;
a = temp;
u = multi(d,(2*d+1)); u = shifting(u,5);
t = multi(b,(2*b+1)); t = shifting(t,5);
c = RC6sub(c,s(2*i+2,:)); c = shifting
(c,-LSB5(t)); c = bitxor(c,u);
a = RC6sub(a,s(2*i+1,:)); a = shifting
(a,-LSB5(u)); a = bitxor(a,t);
end d = RC6sub(d,s(2,:));
b = RC6sub(b,s(1,:)); y(1:4) = a;
y(5:8) = b; y(9:12) = c; y(13:16) = d;

function [xa1,Fs,nbits] = RC6DecCBCAudio(x1, Fs,

nbits,key,r,CO)

%Decrypt Audio using RC6 in CBC mode

%CBC keyi = RC6keygen(key,r); L = length(x1);
y1 = double(x1); y1 = reshape(y1,L/16,16);
plaintextRC6 = zeros(L/16,16);
for i = 1:L/16
RC6enctextss = y1(i,:);
re_plaintext = RC6dec(RC6enctextss,r,keyi);
if i = =1

plaintext = bitxor(re_plaintext,CO);
else
plaintext = bitxor(re_plaintext,y1((i-1),:));
end;
 plaintextRC6(i,:) = plaintext; end;
xa1 = uint8(plaintextRC6(:));

function xa1 = RC6DecCBCImage(x1,key,r,CO)

%Decrypt Image using RC6 in CBC mode

keyi = RC6keygen(key,r);
 App en d i x A 255

%CBC [m,n] = size(x1);

y1 = double(x1);
plaintextRC6 = zeros(m*n/16,16);
RC6enctext1 = reshape(y1',16,m*n/16)'; for i = 1:m*n/16

RC6enctextss = RC6enctext1(i,:);
re_plaintext = RC6dec(RC6enctextss,r,keyi);
if i = =1

plaintext = bitxor(re_plaintext,CO);
else
 plaintext = bitxor(re_plaintext,RC6enctext1
((i-1),:));
end;
plaintextRC6(i,:) = plaintext;

end;
ya1 = reshape(plaintextRC6',n,m)';

xa1 = uint8(ya1);

function xa1 = RC6DecCBCImageC(x,key,r,CO)

%Decrypt Colored Image using RC6 in CBC mode

x1 = x(:,:,1); x2 = x(:,:,2); x3 = x(:,:,3);

y1 = double(x1); y2 = double(x2);
y3 = double(x3); [m,n] = size(x1);
plaintext(:,:,1) = reshape(y1',16,m*n/16)';
plaintext(:,:,2) = reshape(y2',16,m*n/16)';
plaintext(:,:,3) = reshape(y3',16,m*n/16)';

ciphertext = x;
keyi = RC6keygen(key,r);
%CBC
for i = 1:3 plaintextRC6 = plaintext(:,:,i);
 RC6enctext = RC6CBCDec(plaintextRC6,m,n,CO,r,keyi);
 RC6enctext = reshape(RC6enctext',n,m)';
ciphertext(:,:,i) = RC6enctext;
end

xa1 = uint8(ciphertext);

function [xa1, Fs, nbits] = RC6DecCFBAudio(x1, Fs,

nbits,key,r,CO)

%Decrypt Audio using RC6 in CFB mode

256 App en d i x A

%CFB keyi = RC6keygen(key,r); L = length(x1);

y1 = double(x1); y1 = reshape(y1,L/16,16);
plaintextRC6 = zeros(L/16,16);
for i = 1:L/16
RC6enctextss = y1(i,:);
if i = =1 cr1 = RC6enc(CO,r,keyi);
re_plaintext = bitxor(RC6enctextss,cr1);
else
cr1 = RC6enc(y1(i-1,:),r,keyi);
re_plaintext = bitxor(RC6enctextss,cr1);
end;
plaintextRC6(i,:) = re_plaintext;

end;
xa1 = uint8(plaintextRC6(:));

function xa1 = RC6DecCFBImage(x1,key,r,CO)

%Decrypt Image using RC6 in CFB mode

keyi = RC6keygen(key,r);
%CFB [m,n] = size(x1);
y1 = double(x1);
plaintextRC6 = zeros(m*n/16,16);
RC6enctext1 = reshape(y1',16,m*n/16)'; for i = 1:m*n/16
RC6enctextss = RC6enctext1(i,:);
if i = =1 cr1 = RC6enc(CO,r,keyi);

re_plaintext =
bitxor(RC6enctextss,cr1);
else
cr1 = RC6enc(RC6enctext1(i-1,:),r,keyi);

re_plaintext =
bitxor(RC6enctextss,cr1);
end;

plaintextRC6(i,:) = re_plaintext;

end;
ya1 = reshape(plaintextRC6',n,m)';

xa1 = uint8(ya1);

function xa1 = RC6DecCFBImageC(x,key,r,CO)

%Decrypt Colored Image using RC6 in CFB mode
 App en d i x A 257

x1 = x(:,:,1); x2 = x(:,:,2); x3 = x(:,:,3);

y1 = double(x1); y2 = double(x2);
y3 = double(x3); [m,n] = size(x1);
plaintext(:,:,1) = reshape(y1',16,m*n/16)';
plaintext(:,:,2) = reshape(y2',16,m*n/16)';
plaintext(:,:,3) = reshape(y3',16,m*n/16)';​
ciphertext = x;
keyi = RC6keygen(key,r);
%CFB
for i = 1:3 plaintextRC6 = plaintext(:,:,i);
 RC6enctext = RC6CFBDec(plaintextRC6,m,n,CO,r,keyi);
 RC6enctext = reshape(RC6enctext',n,m)';
ciphertext(:,:,i) = RC6enctext;
end

xa1 = uint8(ciphertext);

function [xa1, Fs, nbits] = RC6DecECBAudio(x1, Fs,

nbits,key,r)
%Decrypt Audio using RC6 in ECB mode

%ECB keyi = RC6keygen(key,r); L = length(x1);

y1 = double(x1); y1 = reshape(y1,L/16,16);
plaintextRC6 = zeros(L/16,16);
for i = 1:L/16
RC6enctextss = y1(i,:);
re_plaintext = RC6dec(RC6enctextss,r,keyi);
plaintextRC6(i,:) = re_plaintext;

end;

xa1 = uint8(plaintextRC6(:));

function xa1 = RC6DecECBImage(x1,key,r)

%Decrypt Image using RC6 in ECB mode
keyi = RC6keygen(key,r);
%ECB [m,n] = size(x1);
y1 = double(x1);
plaintextRC6 = zeros(m*n/16,16);
RC6enctext1 = reshape(y1',16,m*n/16)'; for i = 1:m*n/16
RC6enctextss = RC6enctext1(i,:);
re_plaintext = RC6dec(RC6enctextss,r,keyi);
plaintextRC6(i,:) = re_plaintext;

end;
258 App en d i x A

ya1 = reshape(plaintextRC6',n,m)';

xa1 = uint8(ya1);

function xa1 = RC6DecECBImageC(x,key,r)

%Decrypt Colored Image using RC6 in ECB mode

x1 = x(:,:,1); x2 = x(:,:,2); x3 = x(:,:,3);

y1 = double(x1); y2 = double(x2);
y3 = double(x3); [m,n] = size(x1);
plaintext(:,:,1) = reshape(y1',16,m*n/16)';
plaintext(:,:,2) = reshape(y2',16,m*n/16)';
plaintext(:,:,3) = reshape(y3',16,m*n/16)'; ​
ciphertext = x;
keyi = RC6keygen(key,r);
%ECB
for i = 1:3 plaintextRC6 = plaintext(:,:,i);
RC6enctext = RC6ECBDec(plaintextRC6,m,n,r,keyi);
 RC6enctext = reshape(RC6enctext',n,m)';
ciphertext(:,:,i) = RC6enctext;
end xa1 = uint8(ciphertext);

function [xa1, Fs, nbits] = RC6DecOFBAudio(x1, Fs,

nbits,key,r,CO)
%Decrypt Audio using RC6 in OFB mode

%OFB keyi = RC6keygen(key,r); L = length(x1);

y1 = double(x1); y1 = reshape(y1,L/16,16);
plaintextRC6 = zeros(L/16,16);
for i = 1:L/16
RC6enctextss = y1(i,:);
if i = =1
cr = RC6enc (CO,r,keyi);
re_plaintext = bitxor(RC6enctextss,cr);
else cr = RC6enc(cr,r,keyi);
re_plaintext = bitxor(RC6enctextss,cr);
end;
plaintextRC6(i,:) = re_plaintext;
end;
xa1 = uint8(plaintextRC6(:));

function xa1 = RC6DecOFBImage(x1,key,r,CO)

%Decrypt Image using RC6 in OFB mode

keyi = RC6keygen(key,r);
 App en d i x A 259

%OFB [m,n] = size(x1);

y1 = double(x1);
plaintextRC6 = zeros(m*n/16,16);
RC6enctext1 = reshape(y1',16,m*n/16)'; for i = 1:m*n/16
RC6enctextss = RC6enctext1(i,:);
if i = =1
cr = RC6enc (CO,r,keyi);
re_plaintext = bitxor(RC6enctextss,cr);
else cr = RC6enc(cr,r,keyi);
re_plaintext = bitxor(RC6enctextss,cr);
end;

plaintextRC6(i,:) = re_plaintext;

end;
ya1 = reshape(plaintextRC6',n,m)';

xa1 = uint8(ya1);

function xa1 = RC6DecOFBImageC(x,key,r,CO)

%Decrypt Colored Image using RC6 in OFB mode

x1 = x(:,:,1); x2 = x(:,:,2); x3 = x(:,:,3);

y1 = double(x1); y2 = double(x2);
y3 = double(x3); [m,n] = size(x1);
plaintext(:,:,1) = reshape(y1',16,m*n/16)';
plaintext(:,:,2) = reshape(y2',16,m*n/16)';
plaintext(:,:,3) = reshape(y3',16,m*n/16)';
ciphertext = x;
keyi = RC6keygen(key,r);
%OFB
for i = 1:3 plaintextRC6 = plaintext(:,:,i);
 RC6enctext = RC6OFBDec(plaintextRC6,m,n,CO,r,keyi);
 RC6enctext = reshape(RC6enctext',n,m)';
ciphertext(:,:,i) = RC6enctext;
end

xa1 = uint8(ciphertext);

function RC6enctextRC6 = RC6ECB(plaintext1,m,n,r,keyi)

%Encrypt data using RC6 in ECB mode

RC6enctextRC6 = zeros(m*n/16,16);
for i = 1:m*n/16 plaintext = plaintext1(i,:);
RC6enctext = RC6enc(plaintext,r,keyi);
260 App en d i x A

RC6enctextRC6(i,:) = RC6enctext;

end;

function plaintextRC6 =
RC6ECBDec(RC6enctext1,m,n,r,keyi)
%Decrypt Data using RC6 in ECB mode

plaintextRC6 = zeros(m*n/16,16);
for i = 1:m*n/16
RC6enctextss = RC6enctext1(i,:);
re_plaintext = RC6dec(RC6enctextss,r,keyi);
plaintextRC6(i,:) = re_plaintext;

end;

function y = RC6enc(plaintext,round,s)
%RC6 Encryption
a = plaintext(1:4); b = plaintext(5:8);
c = plaintext(9:12); d = plaintext(13:16);
b = add(b,s(1,:)); d = add(d,s(2,:));
for i = 1:round t = multi(b,(2*b+1)); t = shifting(t,5);
u = multi(d,(2*d+1)); u = shifting(u,5);
a = bitxor(a,t'); a = shifting(a,LSB5(u));
a = add(a,s(2*i+1,:)); c = bitxor(c,u');
c = shifting(c,LSB5(t)); c = add(c,s(2*i+2,:));
temp = a;
a = b; b = c; c = d; d = temp;
end a = add(a,s(2*round+3,:)); c =
add(c,s(2*round+4,:)); y(1:4) = a;
y(5:8) = b; y(9:12) = c; y(13:16) = d;

function [xa, Fs, nbits] = RC6EncCBCAudio(x, Fs,

nbits,key,r,CO)
%Encrypt Audio using RC6 in CBC mode
L = length(x);
y = double(x); keyi = RC6keygen(key,r);
plaintext1 = reshape(y,L/16,16);
RC6enctextRC6 = zeros(L/16,16);
%CBC
for i = 1:L/16
plaintext = plaintext1(i,:);
if i = =1
plaintexts = bitxor(plaintext,CO);
else plaintexts = bitxor(plaintext,RC6enctext);
 App en d i x A 2 61

end;
RC6enctext = RC6enc(plaintexts,r,keyi);
RC6enctextRC6(i,:) = RC6enctext;
end;
xa = uint8(RC6enctextRC6(:));

function xa = RC6EncCBCImage(x,key,r,CO)
%Encrypt Image using RC6 in CBC mode
[m,n] = size(x);
y = double(x);
keyi = RC6keygen(key,r);
plaintext1 = reshape(y',16,m*n/16)';
RC6enctextRC6 = zeros(m*n/16,16);
%CBC
for i = 1:m*n/16
plaintext = plaintext1(i,:);
if i = =1
plaintexts = bitxor(plaintext,CO);
else
plaintexts = bitxor(plaintext,RC6enctext);
end;
RC6enctext = RC6enc(plaintexts,r,keyi);
RC6enctextRC6(i,:) = RC6enctext;

end;
ya = reshape(RC6enctextRC6',n,m)';
xa = uint8(ya);

function xa = RC6EncCBCImageC(x,key,r,CO)
%Encrypt Colored Image using RC6 in CBC mode
x1 = x(:,:,1);
x2 = x(:,:,2);
x3 = x(:,:,3);
[m,n] = size(x1);
y1 = double(x1);
y2 = double(x2);
y3 = double(x3);
keyi = RC6keygen(key,r);
plaintext(:,:,1) = reshape(y1',16,m*n/16)';
plaintext(:,:,2) = reshape(y2',16,m*n/16)';
plaintext(:,:,3) = reshape(y3',16,m*n/16)';
RC6enctext = x;
%CBC
for i = 1:3 plaintext1 = plaintext(:,:,i);
RC6enctextRC6 = RC6CBC(plaintext1,m,n,CO,r,keyi);
262 App en d i x A

RC6enctextRC6 = reshape(RC6enctextRC6',n,m)';
RC6enctext(:,:,i) = RC6enctextRC6;
end;
xa = uint8(RC6enctext);

function [xa, Fs, nbits] = RC6EncCFBAudio(x, Fs,

nbits,key,r,CO)
%Encrypt Audio using RC6 in CFB mode

L = length(x);
y = double(x);
keyi = RC6keygen(key,r);
plaintext1 = reshape(y,L/16,16);
RC6enctextRC6 = zeros(L/16,16);
% CFB
for i = 1:L/16
plaintext = plaintext1(i,:);
if i = =1
cr1 = RC6enc(CO,r,keyi);

RC6enctext = bitxor(plaintext,cr1);
else
cr1 = RC6enc(RC6enctext,r,keyi);

RC6enctext = bitxor(plaintext,cr1);

end;

RC6enctextRC6(i,:) = RC6enctext;

end;

xa = uint8(RC6enctextRC6(:));

function xa = RC6EncCFBImage(x,key,r,CO)
%Encrypt Image using RC6 in CFB mode
[m,n] = size(x);
y = double(x);
keyi = RC6keygen(key,r);
plaintext1 = reshape(y',16,m*n/16)';
RC6enctextRC6 = zeros(m*n/16,16);
% CFB
for i = 1:m*n/16
plaintext = plaintext1(i,:);
if i = =1
 App en d i x A 263

cr1 = RC6enc(CO,r,keyi);

RC6enctext = bitxor(plaintext,cr1);
else
cr1 = RC6enc(RC6enctext,r,keyi);
RC6enctext = bitxor(plaintext,cr1);

end;

RC6enctextRC6(i,:) = RC6enctext;
end;

ya = reshape(RC6enctextRC6',n,m)';
xa = uint8(ya);

function xa = RC6EncCFBImageC(x,key,r,CO)
%Encrypt Colored Image using RC6 in CFB mode
x1 = x(:,:,1);
x2 = x(:,:,2);
x3 = x(:,:,3);
[m,n] = size(x1);
y1 = double(x1);
y2 = double(x2);
y3 = double(x3);
keyi = RC6keygen(key,r);
plaintext(:,:,1) = reshape(y1',16,m*n/16)';
plaintext(:,:,2) = reshape(y2',16,m*n/16)';
plaintext(:,:,3) = reshape(y3',16,m*n/16)';
RC6enctext = x;
% CFB
for i = 1:3 plaintext1 = plaintext(:,:,i);
 RC6enctextRC6 = RC6CFB(plaintext1,m,n,CO,r,keyi);
 RC6enctextRC6 = reshape(RC6enctextRC6',n,m)';
RC6enctext(:,:,i) = RC6enctextRC6;
end;
xa = uint8(RC6enctext);

function [xa, Fs, nbits] = RC6EncECBAudio(x, Fs,

nbits,key,r)
%Encrypt Audio using RC6 in ECB mode
L = length(x);
y = double(x);
keyi = RC6keygen(key,r);
plaintext1 = reshape(y,L/16,16);
RC6enctextRC6 = zeros(L/16,16);
264 App en d i x A

% ECB
for i = 1:L/16
plaintext = plaintext1(i,:);
RC6enctext = RC6enc(plaintext,r,keyi);
RC6enctextRC6(i,:) = RC6enctext;
end;

xa = uint8(RC6enctextRC6(:));

function xa = RC6EncECBImage(x,key,r)
%Encrypt Image using RC6 in ECB mode
[m,n] = size(x);
y = double(x);
keyi = RC6keygen(key,r);
plaintext1 = reshape(y',16,m*n/16)';
RC6enctextRC6 = zeros(m*n/16,16);
% ECB
for i = 1:m*n/16
plaintext = plaintext1(i,:);
RC6enctext = RC6enc(plaintext,r,keyi);
RC6enctextRC6(i,:) = RC6enctext;

end;
ya = reshape(RC6enctextRC6',n,m)';
xa = uint8(ya);

function xa = RC6EncECBImageC(x,key,r)
%Encrypt Colored Image using RC6 in ECB mode
x1 = x(:,:,1);
x2 = x(:,:,2);
x3 = x(:,:,3);
[m,n] = size(x1);
y1 = double(x1);
y2 = double(x2);
y3 = double(x3);
keyi = RC6keygen(key,r);
plaintext(:,:,1) = reshape(y1',16,m*n/16)';
plaintext(:,:,2) = reshape(y2',16,m*n/16)';
plaintext(:,:,3) = reshape(y3',16,m*n/16)';
RC6enctext = x;
% ECB
for i = 1:3 plaintext1 = plaintext(:,:,i);
RC6enctextRC6 = RC6ECB(plaintext1,m,n,r,keyi);
RC6enctextRC6 = reshape(RC6enctextRC6',n,m)';
RC6enctext(:,:,i) = RC6enctextRC6;
 App en d i x A 265

end;
xa = uint8(RC6enctext);

function [xa, Fs, nbits] = RC6EncOFBAudio(x, Fs,

nbits,key,r,CO)
%Encrypt Audio using RC6 in OFB mode
L = length(x);
y = double(x);
keyi = RC6keygen(key,r);
plaintext1 = reshape(y,L/16,16);
RC6enctextRC6 = zeros(L/16,16);
% OFB
for i = 1:L/16

plaintext = plaintext1(i,:);
if i == 1
cr = RC6enc(CO,r,keyi);
else
cr = RC6enc(cr,r,keyi);
end;
RC6enctext = bitxor(plaintext,cr);
RC6enctextRC6(i,:) = RC6enctext;
end;
xa = uint8(RC6enctextRC6(:));

function xa = RC6EncOFBImage(x,key,r,CO)
%Encrypt Image using RC6 in OFB mode
[m,n] = size(x);
y = double(x);
keyi = RC6keygen(key,r);
plaintext1 = reshape(y',16,m*n/16)';
RC6enctextRC6 = zeros(m*n/16,16);
% OFB
for i = 1:m*n/16
plaintext = plaintext1(i,:);

if i == 1
cr = RC6enc(CO,r,keyi);
else
cr = RC6enc(cr,r,keyi);

end;
RC6enctext = bitxor(plaintext,cr);
RC6enctextRC6(i,:) = RC6enctext;

end;
266 App en d i x A

ya = reshape(RC6enctextRC6',n,m)';
xa = uint8(ya);

function xa = RC6EncOFBImageC(x,key,r,CO)
%Encrypt Colored Image using RC6 in OFB mode
x1 = x(:,:,1);
x2 = x(:,:,2);
x3 = x(:,:,3);
[m,n] = size(x1);
y1 = double(x1);
y2 = double(x2);
y3 = double(x3);
keyi = RC6keygen(key,r);
plaintext(:,:,1) = reshape(y1',16,m*n/16)';
plaintext(:,:,2) = reshape(y2',16,m*n/16)';
plaintext(:,:,3) = reshape(y3',16,m*n/16)';
RC6enctext = x;
% OFB
for i = 1:3 plaintext1 = plaintext(:,:,i);
 RC6enctextRC6 = RC6OFB(plaintext1,m,n,CO,r,keyi);
RC6enctextRC6 = reshape(RC6enctextRC6',n,m)';
RC6enctext(:,:,i) = RC6enctextRC6;
end;
xa = uint8(RC6enctext);

function s = keygen(key,r)
%RC6 Key Generation
p = [99 81 225 91];
q = [185 121 55 158];
s(1,:) = p;
for i = 2:2*r+4
s(i,:) = add(s(i-1,:),q);
end; i = 1;
a = zeros(1,4);
b = zeros(1,4);
v = 3*(2*r+4);
for h = 1:v
s(i,:) = add(s(i,:),a);
s(i,:) = add(s(i,:),b);
s(i,:) = shifting(s(i,:),3);
a = s(i,:);
key = add(key,a);
key = add(key,b);
key = shifting(key,LSB5(add(a,b)));
b = key;
 App en d i x A 267

if i == 2*r+4
i = 1;
end
end

function RC6enctextRC6 = OFB(plaintext1,m,n,CO,r,keyi)

%Encrypt data using RC6 in OFB mode
RC6enctextRC6 = zeros(m*n/16,16);
for i = 1:m*n/16
plaintext = plaintext1(i,:);

if i == 1
cr = RC6enc(CO,r,keyi);

else
cr = RC6enc(cr,r,keyi);

end;

RC
 6enctext = bitxor(plaintext,cr); RC6enctextRC6(i,:)
= RC6enctext;

end;

fu
nction plaintextRC6 = RC6OFBDec(RC6enctext1,m,n,CO,r,​
keyi)
%Decrypt data using RC6 in OFB mode
plaintextRC6 = zeros(m*n/16,16);
for i = 1:m*n/16
RC6enctextss = RC6enctext1(i,:);
if i = =1
cr = RC6enc (CO,r,keyi);
re_plaintext = bitxor(RC6enctextss,cr);
else

cr = RC6enc(cr,r,keyi);
re_plaintext = bitxor(RC6enctextss,cr);
end;

plaintextRC6(i,:) = re_plaintext;

end;

function y = RC6sub(x,w)
%z = a-a1mod 2^32
268 App en d i x A

a = frombase256(x);
a1 = frombase256(w);
z = mod(a-a1,2^32);
y = tobase256(z);

function out = shifting (w,n)

% shifting w by n
y = dec2bin(w,8);
y = rot90 (y); y = y(:);
y = circshift(y,n);
y = reshape(y,8,4)';
out = bin2dec(y(:,end:-1:1));

function y = shiftleft(key,round)
% RC6 Shifting
if (round = =1||round = =2||round = =9||round = =16)
y = circshift(key,1); else
y = circshift(key,2); end

function y = tobase256(x)
%convert Hex to 256 bits base
for i = 4:-1:1
y(i) = fix(x/256^(i-1)); x = x-(y(i)
*256^(i-1));
end

function c = add(a,b)

%c = a+b mode 2^8

c = zeros(1,4);
for i = 1:4
c(i) = mod(a(i)+
b(i),2^8); if
(i+1)~ = 5
a(i+1) = a(i+1)+fix((a(i)+b(i))/256);
end;
end;

function y = binvec2decA(x)
%convert binary to decimal
y = 0;
for i = 1:length(x)
y = y+x(i)*2^(i-1);
end;
 App en d i x A 269

function cipherdes1 = cipher(plaindes1,r,keyi)

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%
%this file for the encryption of the plaintext to a
ciphertext using des
% clc
% clear all bin = zeros(8);
for i = 1:8
bin(i,:) = dec2binvecA(plaindes1(i),8);
end;
plaindes = rot90(bin);
plaindes = plaindes(:);
ip = InitialPermutation(plaindes);
left = ip(1:32);
right = ip(33:64);
for round = 1:r
expansion = exp1(right);
xor_one = bitxor(expansion',keyi(round,:));
substitution = DESsub(xor_one);
permutation = p(substitution);
xor_two = bitxor(left,permutation);
if round ~ = r left = right;
right = xor_two;
else

left = xor_two;
end;
end;
y(1:32) = left; y(33:64) = right;

cipherdesBIN1 = finalpermutation(y);

cipherdesBIN1 = reshape(cipherdesBIN1,8,8)';
cipherdes1 = zeros(1,8);
for i = 1:8
 cipherdes1(i) = binvec2decA(cipherdesBIN1​
(i,end:-1:1));
end;

function y = Convert256toHex(x)
%convert 256 base numbers to Hex.
a = length(x);
j = 1;
y = zeros(1,2*a);
for i = 1:a
270 App en d i x A

y(j) = fix(x(i)/16); y(j+1) = x(i)-y(j)*16; j = j+2;

end

function y = dec2binvecA(x,size)
%Convert decimal to binary
y = zeros(1,size);
z = zeros(1,size+1);
z(1) = x;
for i = 1:size
y(i) = mod(z(i),2^i)/2^(i-1);
z(i+1) = z(i)-y(i)*2^(i-1);
end;
% y = y(end:-1:1);

function plaindes1 = decry(cipherdes1,r,keyi)

%DES decryption
bin = zeros(8);
for i = 1:8
bin(i,:) = dec2binvecA(cipherdes1(i),8);
end;
cipherdes = rot90(bin);
cipherdes = cipherdes(:);

ip = InitialPermutation(cipherdes);

left = ip(1:32);
right = ip(33:64);

for round = r:-1:1 expansion = exp1(right);

xor_one = bitxor(expansion',keyi(round,:));
substitution = DESsub(xor_one);
permutation = p(substitution);
xor_two = bitxor(left,permutation);
if round ~ = 1
left = right;
right = xor_two;
else
left = xor_two;
end;
end;

y(1:32) = left;
y(33:64) = right;

plaindesBIN1 = finalpermutation(y);
 App en d i x A 2 71

plaindesBIN1 = reshape(plaindesBIN1,8,8)';
plaindes1 = zeros(1,8);
for i = 1:8
plaindes1(i) = binvec2decA(plaindesBIN1(i,
end:-1:1));
end;

fun
ction DESenctextDES = DESCBC(plaintext1,m,n,CO,r,​
keyi)
%Encrypt data using DES in CBC mode
DESenctextDES = zeros(m*n/8,8);
for i = 1:m*n/8
plaintext = plaintext1(i,:);
if i == 1
plaintexts = bitxor(plaintext,CO);
else
plaintexts = bitxor(plaintext,DESenctext);
end;
DESenctext = cipher(plaintexts,r,keyi);
DESenctextDES(i,:) = DESenctext;

end;

fun
ction plaintextDES = DESCBCDec(DESenctext1,m,n,CO,​
r,keyi)
%Decrypt data using DES in CBC mode

plaintextDES = zeros(m*n/8,8);
for i = 1:m*n/8
DESenctextss = DESenctext1(i,:);
re_plaintext = decry(DESenctextss,r,keyi);
if i == 1
plaintext = bitxor(re_plaintext,CO);
else
 plaintext = bitxor(re_plaintext,DESenctext1
((i-1),:));
end;
plaintextDES(i,:) = plaintext;
end;

fun
ction ciphertextDES = DESCFB(plaintext1,m,n,CO,r,​
keyi)
%Encrypt data using DES in CFB mode

ciphertextDES = zeros(m*n/8,8);
272 App en d i x A

for i = 1:m*n/8
plaintext = plaintext1(i,:);
if i == 1
cr1 = cipher(CO,r,keyi);

ciphertext = bitxor(plaintext,cr1);
else
cr1 = cipher(ciphertext,r,keyi);

ciphertext = bitxor(plaintext,cr1);
end;
ciphertextDES(i,:) = ciphertext;

end;

fun
ction plaintextDES = DESCFBDec(ciphertext1,m,n,CO,​
r,keyi)
%decrypt data using DES in CFB mode

plaintextDES = zeros(m*n/8,8);
for i = 1:m*n/8
ciphertextss = ciphertext1(i,:);
if i = =1
cr1 = cipher(CO,r,keyi);

re_plaintext = bitxor(ciphertextss,cr1);
else
cr1 = cipher(ciphertext1(i-1,:),r,keyi);

re_plaintext = bitxor(ciphertextss,cr1);
end;

plaintextDES(i,:) = re_plaintext;

end;

fun
ction [xa1,Fs,nbits] = DesDecCBCAudio(x1,Fs,nbits,​
key,r,CO)
%Decryption Audio using DES in CBC mode

keyi = DESkeygen(key,r);
%CBC L = length(x1);
y1 = double(x1);
plaintextDES = zeros(L/8,8);
ciphertext1 = reshape(y1,L/8,8);
 App en d i x A 2 73

for i = 1:L/8
ciphertextss = ciphertext1(i,:);
re_plaintext = decry(ciphertextss,r,keyi);
if i == 1
plaintext = bitxor(re_plaintext,CO);
else
 plaintext = bitxor(re_plaintext,ciphertext1
((i-1),:));
end;
plaintextDES(i,:) = plaintext;
end;
xa1 = uint8(plaintextDES(:));

function xa1 = DesDecCBCImage(x1,key,r,CO)

%Decrypt Image using DES in CBC mode

keyi = DESkeygen(key,r);
%CBC
[m,n] = size(x1);
y1 = double(x1);
plaintextDES = zeros(m*n/8,8);
ciphertext1 = reshape(y1',8,m*n/8)';

for i = 1:m*n/8
ciphertextss = ciphertext1(i,:);
re_plaintext = decry(ciphertextss,r,keyi);
if i == 1

plaintext = bitxor(re_plaintext,CO);
else
 plaintext = bitxor(re_plaintext,ciphertext1
((i-1),:));
end;
plaintextDES(i,:) = plaintext;

end;
ya1 = reshape(plaintextDES',n,m)';

xa1 = uint8(ya1);

function xa1 = DesDecCBCImageC(x,key,r,CO)

%Decrypt Colored Image using DES in CBC mode

keyi = DESkeygen(key,r);
%CBC
2 74 App en d i x A

x1 = x(:,:,1);
x2 = x(:,:,2);
x3 = x(:,:,3);
y1 = double(x1);
y2 = double(x2);
y3 = double(x3);
[m,n] = size(x1);
plaintext(:,:,1) = reshape(y1',8,m*n/8)';

plaintext(:,:,2) = reshape(y2',8,m*n/8)';
plaintext(:,:,3) = reshape(y3',8,m*n/8)';
ciphertext = x;
for i = 1:3
plaintextDES = plaintext(:,:,i);
DESenctext = DESCBCDec(plaintextDES,m,n,CO,r,keyi);
DESenctext = reshape(DESenctext',n,m)';
ciphertext(:,:,i) = DESenctext;

end;
a1 = uint8(ciphertext);

fun
ction [xa1,Fs,nbits] = DesDecCFBAudio(x1,Fs,nbits,​
key,r,CO)
%Decrypt Audio using DES in CFB mode

keyi = DESkeygen(key,r);
%CFB L = length(x1);
y1 = double(x1);
plaintextDES = zeros(L/8,8);
ciphertext1 = reshape(y1,L/8,8);
for i = 1:L/8
ciphertextss = ciphertext1(i,:);
if i == 1
cr1 = cipher(CO,r,keyi);

re_plaintext = bitxor(ciphertextss,cr1);
else
cr1 = cipher(ciphertext1(i-1,:),r,keyi);

re_plaintext = bitxor(ciphertextss,cr1);
end;

plaintextDES(i,:) = re_plaintext;

end;
 App en d i x A 275

xa1 = uint8(plaintextDES(:));

function xa1 = DesDecCFBImage(x1,key,r,CO)

%Decrypt Image using DES in CFB mode

keyi = DESkeygen(key,r);
%CFB
[m,n] = size(x1);
y1 = double(x1);
plaintextDES = zeros(m*n/8,8);
ciphertext1 = reshape(y1',8,m*n/8)';
for i = 1:m*n/8
ciphertextss = ciphertext1(i,:);
if i == 1
cr1 = cipher(CO,r,keyi);

re_plaintext = bitxor(ciphertextss,cr1);
else
cr1 = cipher(ciphertext1(i-1,:),r,keyi);

re_plaintext = bitxor(ciphertextss,cr1);
end;
plaintextDES(i,:) = re_plaintext;
end;
ya1 = reshape(plaintextDES',n,m)';

xa1 = uint8(ya1);

function xa1 = DesDecCFBImageC(x,key,r,CO)

%Decrypt Colored Image using DES in CFB mode

keyi = DESkeygen(key,r);
%CFB
x1 = x(:,:,1);
x2 = x(:,:,2);
x3 = x(:,:,3);
y1 = double(x1);
y2 = double(x2);
y3 = double(x3);
[m,n] = size(x1);
plaintext(:,:,1) = reshape(y1',8,m*n/8)';
plaintext(:,:,2) = reshape(y2',8,m*n/8)';
plaintext(:,:,3) = reshape(y3',8,m*n/8)';
ciphertext = x;
for i = 1:3 plaintextDES = plaintext(:,:,i);
276 App en d i x A

 DESenctext = DESCFBDec(plaintextDES,m,n,CO,r,keyi);
DESenctext = reshape(DESenctext',n,m)';
ciphertext(:,:,i) = DESenctext;
end

xa1 = uint8(ciphertext);

fun
ction [xa1,Fs,nbits] = DesDecECBAudio(x1,Fs,nbits,​
key,r)
%Decrypt Audio using DES in ECB mode

keyi = DESkeygen(key,r);
%ECB L = length(x1);
y1 = double(x1);
plaintextDES = zeros(L/8,8);
ciphertext1 = reshape(y1,L/8,8);
for i = 1:L/8 ciphertextss = ciphertext1(i,:);
re_plaintext = decry(ciphertextss,r,keyi);
plaintextDES(i,:) = re_plaintext;

end;
xa1 = uint8(plaintextDES(:));

function xa1 = DesDecECBImage(x1,key,r)

%Decrypt Image using DES in ECB mode

keyi = DESkeygen(key,r);
%ECB
[m,n] = size(x1);
y1 = double(x1);
plaintextDES = zeros(m*n/8,8);
ciphertext1 = reshape(y1',8,m*n/8)';
for i = 1:m*n/8
ciphertextss = ciphertext1(i,:);
re_plaintext = decry(ciphertextss,r,keyi);
plaintextDES(i,:) = re_plaintext;
end;
ya1 = reshape(plaintextDES',n,m)';
xa1 = uint8(ya1);

function xa1 = DesDecECBImageC(x,key,r)

%Decrypt Colored Image using DES in ECB mode

keyi = DESkeygen(key,r);
%ECB x1 = x(:,:,1);
 App en d i x A 277

x2 = x(:,:,2);
x3 = x(:,:,3);
y1 = double(x1);
y2 = double(x2);
y3 = double(x3);
[m,n] = size(x1);
plaintext(:,:,1) = reshape(y1',8,m*n/8)';
plaintext(:,:,2) = reshape(y2',8,m*n/8)';
plaintext(:,:,3) = reshape(y3',8,m*n/8)';
ciphertext = x;
for i = 1:3 plaintextDES = plaintext(:,:,i);
DESenctext = DESECBDec(plaintextDES,m,n,r,keyi);
DESenctext = reshape(DESenctext',n,m)';
ciphertext(:,:,i) = DESenctext;
end
xa1 = uint8(ciphertext);

fun
ction [xa1,Fs,nbits] = DesDecOFBAudio(x1,Fs,nbits,​
key,r,CO)
%Decrypt Audio using DES in OFB mode

keyi = DESkeygen(key,r);
%OFB L = length(x1);
y1 = double(x1);
plaintextDES = zeros(L/8,8);
ciphertext1 = reshape(y1,L/8,8);
for i = 1:L/8
ciphertextss = ciphertext1(i,:);
if i = =1
cr = cipher(CO,r,keyi);
re_plaintext = bitxor(ciphertextss,cr);
else cr = cipher(cr,r,keyi);
re_plaintext = bitxor(ciphertextss,cr);
end;

plaintextDES(i,:) = re_plaintext;

end;
xa1 = uint8(plaintextDES(:));

function xa1 = DesDecOFBImage(x1,key,r,CO)

%Decrypt Image using DES in OFB mode

keyi = DESkeygen(key,r);
%OFB [m,n] = size(x1);
2 78 App en d i x A

y1 = double(x1);
plaintextDES = zeros(m*n/8,8);
ciphertext1 = reshape(y1',8,m*n/8)';
for i = 1:m*n/8
ciphertextss = ciphertext1(i,:);
if i == 1 cr = cipher(CO,r,keyi);
re_plaintext = bitxor(ciphertextss,cr);
else cr = cipher(cr,r,keyi);
re_plaintext = bitxor(ciphertextss,cr);
end;
plaintextDES(i,:) = re_plaintext;
end;
ya1 = reshape(plaintextDES',n,m)';
xa1 = uint8(ya1);

function xa1 = DesDecOFBImageC(x,key,r,CO)

%Decrypt Colored Image using DES in OFB mode

keyi = DESkeygen(key,r);
%OFB x1 = x(:,:,1);
x2 = x(:,:,2);
x3 = x(:,:,3);
y1 = double(x1);
y2 = double(x2);
y3 = double(x3);
[m,n] = size(x1);
plaintext(:,:,1) = reshape(y1',8,m*n/8)';
plaintext(:,:,2) = reshape(y2',8,m*n/8)';
plaintext(:,:,3) = reshape(y3',8,m*n/8)';
ciphertext = x;
for i = 1:3 plaintextDES = plaintext(:,:,i);
DESenctext = DESOFBDec(plaintextDES,m,n,CO,r,​
keyi);
DESenctext = reshape(DESenctext',n,m)';
ciphertext(:,:,i) = DESenctext;
end

xa1 = uint8(ciphertext);

function DESenctextDES = DESECB(plaintext1,m,n,r,keyi)

%Encrypt Data using DES in ECB mode

DESenctextDES = zeros(m*n/8,8);
for i = 1:m*n/8
plaintext = plaintext1(i,:);
DESenctext = cipher(plaintext,r,keyi);
 App en d i x A 2 79

DESenctextDES(i,:) = DESenctext;

end;

fun
ction plaintextDES = DESECBDec(DESenctext1,m,n,r,​
keyi)
%Decrypt data using DES in ECB mode

plaintextDES = zeros(m*n/8,8);
for i = 1:m*n/8
DESenctextss = DESenctext1(i,:);
re_plaintext = decry(DESenctextss,r,keyi);
plaintextDES(i,:) = re_plaintext;

end;

fun
ction [xa,Fs,nbits] = DesEncCBCAudio(x,Fs,nbits,​
key,r,CO)
%Encrypt Audio using DES in CBC mode

L = length(x);
y = double(x);
keyi = DESkeygen(key,r);
plaintext1 = reshape(y,L/8,8);
ciphertextDES = zeros(L/8,8);
%CBC
for i = 1:L/8
plaintext = plaintext1(i,:);
if i == 1
plaintexts = bitxor(plaintext,CO);
else
plaintexts = bitxor(plaintext,ciphertext);
end;
ciphertext = cipher(plaintexts,r,keyi);
ciphertextDES(i,:) = ciphertext;
end;
xa = uint8(ciphertextDES(:));

function xa = DesEncCBCImage(x,key,r,CO)
%Encrypt Image using DES in CBC mode
[m,n] = size(x);
y = double(x);
keyi = DESkeygen(key,r);
plaintext1 = reshape(y',8,m*n/8)';
ciphertextDES = zeros(m*n/8,8);
280 App en d i x A

%CBC
for i = 1:m*n/8
plaintext = plaintext1(i,:);
if i = =1
plaintexts = bitxor(plaintext,CO);
else
plaintexts = bitxor(plaintext,ciphertext);
end;
ciphertext = cipher(plaintexts,r,keyi);
ciphertextDES(i,:) = ciphertext;

end;
ya = reshape(ciphertextDES',n,m)';
xa = uint8(ya);

function xa = DesEncCBCImageC(x,key,r,CO)
%Encrypt Colored Image using DES in CBC mode
x1 = x(:,:,1);
x2 = x(:,:,2);
x3 = x(:,:,3);
[m,n] = size(x1);
y1 = double(x1);
y2 = double(x2);
y3 = double(x3);
keyi = DESkeygen(key,r);
plaintext(:,:,1) = reshape(y1',8,m*n/8)';
plaintext(:,:,2) = reshape(y2',8,m*n/8)';
plaintext(:,:,3) = reshape(y3',8,m*n/8)';
DESenctext = x;
%CBC
for i = 1:3 plaintext1 = plaintext(:,:,i);
DESenctextDES = DESCBC(plaintext1,m,n,CO,r,keyi);
DESenctextDES = reshape(DESenctextDES',n,m)';
DESenctext(:,:,i) = DESenctextDES;
end;
xa = uint8(DESenctext);

fun
ction [xa,Fs,nbits] = DesEncCFBAudio(x,Fs,nbits,key,​
r,CO)
%Encrypt Audio using DES in CFB mode
L = length(x);
y = double(x);
keyi = DESkeygen(key,r);
plaintext1 = reshape(y,L/8,8);
ciphertextDES = zeros(L/8,8);
 App en d i x A 2 81

% CFB
for i = 1:L/8
plaintext = plaintext1(i,:);
if i = =1
cr1 = cipher(CO,r,keyi);
ciphertext = bitxor(plaintext,cr1);
else cr1 = cipher(ciphertext,r,keyi);
ciphertext = bitxor(plaintext,cr1);
end;
ciphertextDES(i,:) = ciphertext;
end;
xa = uint8(ciphertextDES(:));

function xa = DesEncCFBimage(x,key,r,CO)
%Encrypt Image using DES in CFB mode
[m,n] = size(x);
y = double(x);
keyi = DESkeygen(key,r);
plaintext1 = reshape(y',8,m*n/8)';
ciphertextDES = zeros(m*n/8,8);
% CFB
for i = 1:m*n/8
plaintext = plaintext1(i,:);
if i = =1
cr1 = cipher(CO,r,keyi);
ciphertext = bitxor(plaintext,cr1);
else cr1 = cipher(ciphertext,r,keyi);
ciphertext = bitxor(plaintext,cr1);
end;
ciphertextDES(i,:) = ciphertext;
end;
ya = reshape(ciphertextDES',n,m)';

xa = uint8(ya);

function xa = DesEncCFBImageC(x,key,r,CO)
%Encrypt Colored Image using DES in CFB mode

x1 = x(:,:,1); x2 = x(:,:,2); x3 = x(:,:,3);

[m,n] = size(x1); y1 = double(x1);
y2 = double(x2); y3 = double(x3);
keyi = DESkeygen(key,r);
plaintext(:,:,1) = reshape(y1',8,m*n/8)';
plaintext(:,:,2) = reshape(y2',8,m*n/8)';
282 App en d i x A

plaintext(:,:,3) = reshape(y3',8,m*n/8)';
DESenctext = x;
% CFB
for i = 1:3 plaintext1 = plaintext(:,:,i);
DESenctextDES = DESCFB(plaintext1,m,n,CO,r,keyi);
DESenctextDES = reshape(DESenctextDES',n,m)';
DESenctext(:,:,i) = DESenctextDES;
end;
xa = uint8(DESenctext);

function [xa,Fs,nbits] = DesEncECBAudio(x,Fs,nbits,​

key,r)
%Encrypt Audio using DES in ECB mode
L = length(x);
y = double(x); keyi = DESkeygen(key,r);
plaintext1 = reshape(y,L/8,8);
ciphertextDES = zeros(L/8,8);
% ECB
for i = 1:L/8 plaintext = plaintext1(i,:);
ciphertext = cipher(plaintext,r,keyi);
ciphertextDES(i,:) = ciphertext;
end;
xa = uint8(ciphertextDES(:));

function xa = DesEncECBImage(x,key,r)
%Encrypt Image using DES in ECB mode

[m,n] = size(x);
y = double(x); keyi = DESkeygen(key,r);
plaintext1 = reshape(y',8,m*n/8)';
ciphertextDES = zeros(m*n/8,8);
% ECB
for i = 1:m*n/8 plaintext = plaintext1(i,:);
ciphertext = cipher(plaintext,r,keyi);
ciphertextDES(i,:) = ciphertext;

end; ya = reshape(ciphertextDES',n,m)'; xa = uint8(ya);

function xa = DesEncECBImageC(x,key,r)
%Encrypt Colored Image using DES in ECB mode
x1 = x(:,:,1); x2 = x(:,:,2); x3 = x(:,:,3);
[m,n] = size(x1); y1 = double(x1);
y2 = double(x2);
y3 = double(x3); keyi = DESkeygen(key,r);
plaintext(:,:,1) = reshape(y1',8,m*n/8)';
 App en d i x A 283

plaintext(:,:,2) = reshape(y2',8,m*n/8)';
plaintext(:,:,3) = reshape(y3',8,m*n/8)';

DESenctext = x;

for i = 1:3 plaintext1 = plaintext(:,:,i);

DESenctextDES = DESECB(plaintext1,m,n,r,keyi);
DESenctextDES = reshape(DESenctextDES',n,m)';
DESenctext(:,:,i) = DESenctextDES;
end;
xa = uint8(DESenctext);

fun
ction [xa,Fs,nbits] = DesEncOFBAudio(x,Fs,nbits,​
key,r,CO)
%Encrypt Audio using DES in OFB mode
L = length(x);
y = double(x); keyi = DESkeygen(key,r);
plaintext1 = reshape(y,L/8,8);
ciphertextDES = zeros(L/8,8);
% OFB
for i = 1:L/8 plaintext = plaintext1(i,:); if i == 1
cr = cipher(CO,r,keyi);
else cr = cipher(cr,r,keyi);
end; ciphertext = bitxor(plaintext,cr);
ciphertextDES(i,:) = ciphertext;
end;
xa = uint8(ciphertextDES(:));

function xa = DesEncOFBImage(x,key,r,CO)
%Encrypt Image using DES in OFB mode

[m,n] = size(x);
y = double(x); keyi = DESkeygen(key,r);
plaintext1 = reshape(y',8,m*n/8)';
ciphertextDES = zeros(m*n/8,8);
% OFB
for i = 1:m*n/8 plaintext = plaintext1(i,:); if i == 1
cr = cipher(CO,r,keyi);
else cr = cipher(cr,r,keyi);
end; ciphertext = bitxor(plaintext,cr);
ciphertextDES(i,:) = ciphertext;
end; ya = reshape(ciphertextDES',n,m)'; xa = uint8(ya);

function xa = DesEncOFBImageC(x,key,r,CO)
%Encrypt Colored Image using DES in OFB mode
28 4 App en d i x A

x1 = x(:,:,1);
x2 = x(:,:,2);
x3 = x(:,:,3);
[m,n] = size(x1);
y1 = double(x1);
y2 = double(x2);
y3 = double(x3);
keyi = DESkeygen(key,r);
plaintext(:,:,1) = reshape(y1',8,m*n/8)';
plaintext(:,:,2) = reshape(y2',8,m*n/8)';
plaintext(:,:,3) = reshape(y3',8,m*n/8)';
DESenctext = x;
% OFB
for i = 1:3 plaintext1 = plaintext(:,:,i);
 DESenctextDES = DESOFB(plaintext1,m,n,CO,r,keyi);
DESenctextDES = reshape(DESenctextDES',n,m)';
DESenctext(:,:,i) = DESenctextDES;
end;
xa = uint8(DESenctext);

function key1 = DESkeygen(Mainkey,r)

%DES Key Generation
% Inserting master key
% clc
% clear all

% Mainkey = [
'A';'A';'B';'B';'0';'9';'1';'8';'2';'7';'
3';'6';'C';'C';'D';'D'];
key2 = Convert256toHex(Mainkey);
key = zeros(r,4); for i = 1:r
key(i,:) = dec2binvecA(key2(i),4); end
key = rot90(key);
key = key(:);

a = [
57 49 41 33 25 17 9 1 58 50 42 34 26 18 10 2 59
51 43 35 27 19 11 3 60 52 44 36 63 55 47 39 31
23 15 7 62 54 46 38 30 22 14 6 61 53 45 37 29 21
13 5 28 20 12 4];
b = [
14 17 11 24 1 5 3 28 15 6 21 10 23 19 12 4 26 8
16 7 27 20 13 2 41 52 31 37 47 55 30 40 51 45 33
48 44 49 39 56 34 53 46 42 50 36 29 32];
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
output1 = key(a);
key1 = zeros(r,48);
left = output1(28:-1:1);
 App en d i x A 285

right = output1(56:-1:29);
for i = 1:r
left = shiftleft(left,i);
right = shiftleft(right,i);
output(28:-1:1) = left;
output(56:-1:29) = right;
key1(i,:) = output(b);
end

fun
ction ciphertextDES = DESOFB(plaintext1,m,n,CO,r,​
keyi)
%Encrypt data using DES in OFB mode

ciphertextDES = zeros(m*n/8,8);
for i = 1:m*n/8
plaintext = plaintext1(i,:);

if i == 1
cr = cipher(CO,r,keyi);

else
cr = cipher(cr,r,keyi);

end;
ciphertext = bitxor(plaintext,cr);
ciphertextDES(i,:) = ciphertext;

end;

fun
ction plaintextDES = DESOFBDec(ciphertext1,m,n,CO,​
r,keyi)
%Decrypt data using DES in OFB mode

plaintextDES = zeros(m*n/8,8);
for i = 1:m*n/8
ciphertextss = ciphertext1(i,:);
if i = =1
cr = cipher (CO,r,keyi);
re_plaintext = bitxor(ciphertextss,cr);
else cr = cipher(cr,r,keyi);
re_plaintext = bitxor(ciphertextss,cr);
end;

plaintextDES(i,:) = re_plaintext;

end;
286 App en d i x A

function y = sub(in)
%DES substitution
in1 = reshape(in',6,8)';
x = zeros(1,8);
for i = 1:8
x(i) = binvec2decA(in1(i,end:-1:1));
end;
s = 
[14 0 4 15 13 7 1 4 2 14 15 2 11 13 8 1 3 10 10 6 6
12 12 11 5 9 9 5 0 3 7 8 4 15 1 12 14 8 8 2 13 4 6
9 2 1 11 7 15 5 12 11 9 3 7 14 3 10 10 0 5 6 0 13;
 15 3 1 13 8 4 14 7 6 15 11 2 3 8 4 14 9 12 7 0 2 1
13 10 12 6 0 9 5 11 10 5 0 13 14 8 7 10 11 1 10 3 4
15 13 4 1 2 5 11 8 6 12 7 6 12 9 0 3 5 2 14 15 9;
 10 13 0 7 9 0 14 9 6 3 3 4 15 6 5 10 1 2 13 8 12 5
7 14 11 12 4 11 2 15 8 1 13 1 6 10 4 13 9 0 8 6 15
9 3 8 0 7 11 4 1 5 2 14 12 3 5 11 10 5 14 2 7 12;
 7 13 13 8 14 11 3 5 0 6 6 15 9 0 10 3 1 4 2 7 8 2 5
12 11 1 12 10 4 14 15 9 10 3 6 15 9 0 0 6 12 10 11
1 7 13 13 8 15 9 1 4 3 5 14 11 5 12 2 7 8 2 4 14;
 2 14 12 11 4 2 1 12 7 4 10 7 11 13 6 1 8 5 5 0 3
15 15 10 13 3 0 9 14 8 9 6 4 11 2 8 1 12 11 7
10 1 13 14 7 2 8 13 15 6 9 15 12 0 5 9 6 10 3 4
0 5 14 3;
 12 10 1 15 10 4 15 2 9 7 2 12 6 9 8 5 0 6 13 1 3 13
4 14 14 0 7 11 5 3 11 8 9 4 14 3 15 2 5 12 2 9 8 5
12 15 3 10 7 11 0 14 4 1 10 7 1 6 13 0 11 8 6 13;
 4 13 11 0 2 11 14 7 15 4 0 9 8 1 13 10 3 14 12 3 9
5 7 12 5 2 10 15 6 8 1 6 1 6 4 11 11 13 13 8 12 1 3
4 7 10 14 7 10 9 15 5 6 0 8 15 0 14 5 2 9 3 2 12;
 13 1 2 15 8 13 4 8 6 10 15 3 11 7 1 4 10 12 9 5 3 6
14 11 5 0 0 14 12 9 7 2 7 2 11 1 4 14 1 7 9 4 12 10
14 8 2 13 0 15 6 12 10 9 13 0 15 3 3 5 5 6 8 11];
out1 = zeros(1,8);
for i = 1:8
out1(i) = s(i,x(i)+1);
end
out = zeros(8,4);
for i = 1:8
out(i,:) = dec2binvecA(out1(i),4);
end
out = rot90(out);
y = out(:);

function expansion = exp1(right)

%DES expansion
 App en d i x A 287

a = [
32,1,2,3,4,5,4,5,6,7,8,9,8,9,10,11,12,13,12,13,14,
15,16,17,16,17,18,19,20,21,20,21,22,23,24,25,24,25,
26,27,28,29,28,29,30,31,32,1];
expansion = right(a);

function y = finalpermutation(x)
%DES final permutation

a = [
40 8 48 16 56 24 64 32 39 7 47 15 55 23 63 31 38
6 46 14 54 22 62 30 37 5 45 13 53 21 61 29 36 4 44
12 52 20 60 28 35 3 43 11 51 19 59 27 34 2 42 10
50 18 58 26 33 1 41 9 49 17 57 25];
y = x(a);

function y = frombase256(x)
%Convert 256 base number to decimal
y = 0;
for i = 1:4
y = y+(x(i)*256^(i-1));
end

function ip = InitialPermutation(plaindes)
%DES Initial Permutation
a = [
58 50 42 34 26 18 10 2 60 52 44 36 28 20 12 4 62
54 46 38 30 22 14 6 64 56 48 40 32 24 16 8 57 49
41 33 25 17 9 1 59 51 43 35 27 19 11 3 61 53 45 37
29 21 13 5 63 55 47 39 31 23 15 7];
ip = plaindes(a);

function y = LSB5(x)
%return the 5 least significant bits of x
y = dec2bin(x(1),8);
u = y(8:-1:3); u = u(end:-1:1);
y = bin2dec(u);

function y = CHaoticCipher(im)
% This function encrypts square image using baker map
im = double(im);
n = 
[10,5,12,5,10,8,14,10,5,12,5,10,8,14,10,5,12,5,10,8,
14,10,5,12,5,10,8,14,10,5,12,5,10,8,14,10,5,12,5,10,
8,14,10,5,12,5,10,8,14,10,5,12,5,10,8,14,];
[pr,pc] = chaomat(n);
pim = chaoperm(im,pr,pc,3,'forward');
y = uint8(pim);
imshow(y);
288 App en d i x A

function [pr,pc] = chaomat(n)

%
I = sum(n);
k = size(n,2);
for i = 1:k
N(i+1) = 1;
for j = 1:i
N(i+1) = N(i+1)+n(j);
end
end

N(1) = 1;

for cb = 1:k
for rb = 1:n(cb)
rbstartcol(rb) = mod((rb-1)*I,n(cb));
rbendcol(rb) = mod((rb*I-1),n(cb));
rbstartrow(rb) = fix(((rb-1)*I)/n(cb));
rbendrow(rb) = fix((rb*I-1)/n(cb));
 mincol(rb) = min([rbendcol(rb)+1,rbstartcol(rb)]);
 maxcol(rb) = max([rbendcol(rb),rbstartcol(rb)-1]);
end
for i = 1:I
for j = N(cb):N(cb+1)-1
newindex(i,j-N(cb)+1) = (i-1)*n(cb)+(n(cb)-
j+N(cb)-1);
newindexmod(i,j-N(cb)+1) = mod(newindex(i,j-
N(cb)+1),n(cb));
newindexquotient(i,j-N(cb)+1) = fix(newindex(i,j-
N(cb)+1)/n(cb));
rowblockindex(i,j-N(cb)+1) = fix(newindex(i,j-
N(cb)+1)/I)+1;
end
end
for i = 1:I
for j = 1:n(cb)
for rb = 1:n(cb)
if rowblockindex(i,j) = =rb;
if newindexmod(i,j)>maxcol(rb)
col = rbendrow(rb)-
newindexquotient(i,j)+(n(cb)-1-
newindexmod(i,j))*(rbendrow(rb)-
rbstartrow(rb));
 elseif newindexmod(i,j)> = mincol(rb) &
newindexmod(i,j)< = maxcol(rb)
 App en d i x A 289

if rbstartcol(rb)>rbendcol(rb)
c = 0;
d = -1;
else
c = 1;
d = 1;
end
col = (rbendrow(rb)-
rbstartrow(rb))*(n(cb)-1-
maxcol(rb))+(rbendrow(rb)-newindexquot
ient(i,j)+c)+(maxcol(rb)-
newindexmod(i,j))*(rbendrow(rb)-
rbstartrow(rb)+d);
else%if newindexmod(i,j)< = mincol(rb)
col = I-mincol(rb)*(rbendrow(rb)-
rbstartrow(rb))+(rbendrow(rb)-newindex
quotient(i,j)+1)+(mincol(rb)-1-
newindexmod(i,j))*(rbendrow(rb)-
rbstartrow(rb));
end
row = 1+I-N(cb+1)+rowblockindex(i,j);
end
end
pr(i,j+N(cb)-1) = row;
pc(i,j+N(cb)-1) = col;
end
end
end

function out = chaoperm(im,pr,pc,num,forward)

%
[rows,cols] = size(im);
mat = zeros([rows,cols,num+1]);
mat(:,:,1) = im(:,:);
for loc = 2:num+1
if(strcmp(forward,'forward'))
for i = 1:rows
for j = 1:cols
mat(pr(i,j),pc(i,j),loc) =
mat(i,j,loc-1);
end
end
elseif(strcmp(forward,'backward'))
for i = 1:rows
290 App en d i x A

for j = 1:cols
mat(i,j,loc) = mat(pr(i,j),pc(i,j),loc-1);
end
end
end
end
out = mat(:,:,num+1);

function ID = IDMF(x,y)
%this fuunction(irregular Deviation Measuring Factor) is
based on how much the deviation cased by encryption is
%irregular.
%x:-Original Image
%y:-Encrypted Image
%D:-Maximum Deviation Measuring Factor
x = double(x);
y = double(y);
%first,calculate the difference between each pixel
value before and after
%encryption
D = uint8(abs(x-y));
%calculate the H histogram of the difference
H = imhist(D)
%calculate the average of H
DC = 0;
for i = 1:256
DC = DC+H(i);
end
DC = DC/256;
%subtract the DC from H at every point
for i = 1:256
AC(i) = abs(H(i)-DC);
end
%calculate the sum of AC
ID = 0;
for i = 1:256
ID = ID+AC(i);
End;
%the lower the value, the better the encryption

function D = MDMF(x,y)
%t
his fuunction calculates the Maximum Deviation
Measuring Factor which
%c
alculates the deviation between the original and
encrypted image.
 App en d i x A 2 91

%x:-Original Image
%y:-Encrypted Image
%D:-Maximum Deviation Measuring Factor
%f
irst,calculate the histogram between the original
and encrypted image
x1 = imhist(x);
y1 = imhist(y);
%then calculate the difference between the two
diff = abs(y1-x1);
%then calculate D as follows
D1 = 0;
for i = 2:255
D1 = D1+diff(i);
end;
D2 = (diff(1)+diff(255))/2;
D = D1+D2;
function y = psnr(im1,im2)
%Peak Signal to Noise Ratio
MSEa = 0;
[m,n] = size(im1);
im1 = double(im1);
im2 = double(im2);
for i = 1:m
for j = 1:n
x = double(im1(i,j)-im2(i,j));
x1 = x^2/(m*n);
MSEa = MSEa+x1;
end;
end;
MSEa = double(MSEa);
y = 10*log10((255^2)/MSEa);
Appendix B

%Image with no Encryption over OFDM system + AWGN ....

%Transmitter..........................................
%Data Generation .....................................

f = imread ('Cameraman.tif');
[M,N] = size(f) ;
g1 = im2col(f, [M,N], [M,N], 'distinct');
h1 = dec2bin(double(g1));
[M1,N1] = size(h1) ;
z1 = zeros (M1,N1) ;
for i = 1
M1
for j = 1: N1
z1(i,j) = str2num(h1(i,j));
end;
end;
[R1,T1] = size(z1) ;
zz1 = reshape(z1,R1*T1, 1);
% The transmitted data ...............................
zz = [zz1];
trel = poly2trellis(7,[171 133]);% Trellis
data1 = zz;
zz = convenc(data1,trel);
s1 = length(zz);
para = 128;

293
294 App en d i x B

nd = 6; %number of information OFDM symbol for one

loop
ml = 2; %Modulation level: QPSK
sr = 250000; %symbol rate
br = sr.*ml; %Bit rat per carrier
Ipoint = 8;%Number of over samples
gilen = 32;
flat = 1;
fd = 600;

fftlen = para;
noc = para;
ofdm_length = para*nd*ml; %Total no for one loop

%Dividing the image into blocks ......................

nloops = ceil((length(zz))/ofdm_length);
new_data = nloops*ofdm_length ;
nzeros = new_data - length(zz);
input_data = [zz;zeros(nzeros,1)];
input_data2 = reshape(input_data,ofdm_length,nloops);

%Transmission ON FFT_OFDM ..................... ......

for ebno = [0,2,4,6,8,10];

demodata1 = zeros(ofdm_length,nloops);
for jj = 1: nloops % loop for columns
serdata1 = input_data2(:,jj)';
demodata = fft_channel(serdata1,para,nd,ml,gilen,fftl
en,sr,ebno, br);
 demodata1(:,jj) = demodata(: ); % the
output of ofdm columns
end

%Received image ......................................

[Mr,Nr] = size(demodata1);
% demodata2 = demodata1(: );
yy = reshape (demodata1,Mr*Nr,1);%
part1 = yy(1: s1);
yy11 = part1;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%DECODING ...........
yy11 = vitdec(yy11',trel,1,'term','hard');% Decode.

yy1 = reshape(yy11,[R1,T1]);
for i = 1: M1
 App en d i x B 295

for j = 1: N1
zn1(i,j) = num2str(yy1(i,j));
end;
end;
hn1 = bin2dec(zn1);
gn1 = col2im(hn1, [M,N], [M,N], 'distinct');
%The Error between Trans ............................
output_image = gn1/255;

output_image1 = medfilt2(output_image);

%The Error between Trans .............................

MSE1 = s
um(sum((double(f)/255-output_image).^2))/
prod(size(f));
PSNR = 10*log(1/MSE1)/log(10)

MSE11 = s
um(sum((double(f)/255-output_image1).^2))/
prod(size(f));
PSNR1 = 10*log(1/MSE11)/log(10);

end

%Transmission ON DCT_OFDM .............................

for ebno = [0,2,4,6,8,10];

demodata1 = zeros(ofdm_length,nloops);
for jj = 1: nloops % loop for columns
serdata1 = input_data2(:,jj)';
demodata = 
dct_channel(serdata1,para,nd,ml,gilen,f
ftlen,sr,ebno, br);
demodata1(:,jj) = 
demodata(: ); % the output of
ofdm columns
end

%Received image ......................................

[Mr,Nr] = size(demodata1);
% demodata2 = demodata1(: );
yy = reshape (demodata1,Mr*Nr,1);%
part1 = yy(1: s1);
yy11 = part1;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%DECODING ...........
yy11 = vitdec(yy11',trel,1,'term','hard');% Decode.

yy1 = reshape(yy11,[R1,T1]);
296 App en d i x B

for i = 1: M1
for j = 1: N1
zn1(i,j) = num2str(yy1(i,j));
end;
end;
hn1 = bin2dec(zn1);
gn1 = col2im(hn1, [M,N], [M,N], 'distinct');
%The Error between Trans .......................
output_image = gn1/255;

output_image1 = medfilt2(output_image);

%The Error between Trans .............................

MSE1 = s
um(sum((double(f)/255-output_image).^2))/
prod(size(f));
PSNR = 10*log(1/MSE1)/log(10)
MSE11 = sum(sum((double(f)/255-output_image1).^2))/
prod(size(f));
PSNR1 = 10*log(1/MSE11)/log(10);

end
%Transmission ON DWT_OFDM ............................

for ebno = [0,2,4,6,8,10];

demodata1 = zeros(ofdm_length,nloops);
for jj = 1: nloops % loop for columns
serdata1 = input_data2(:,jj)';
demodata = 
dwt_channel(serdata1,para,nd,ml,gilen
,fftlen,sr,ebno, br);
demodata1(:,jj) = 
demodata(: ); % the
output of ofdm
columns
end

%Received image ......................................

[Mr,Nr] = size(demodata1);
% demodata2 = demodata1(: );
yy = reshape (demodata1,Mr*Nr,1);%
part1 = yy(1: s1);
yy11 = part1;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%DECODING ...........
yy11 = vitdec(yy11',trel,1,'term','hard');% Decode.

yy1 = reshape(yy11,[R1,T1]);
 App en d i x B 297

for i = 1: M1
for j = 1: N1
zn1(i,j) = num2str(yy1(i,j));
end;
end;
hn1 = bin2dec(zn1);
gn1 = col2im(hn1, [M,N], [M,N], 'distinct');
%The Error between Trans ............................
output_image = gn1/255;

output_image1 = medfilt2(output_image);

%The Error between Trans .............................

MSE1 = s
um(sum((double(f)/255-output_image).^2))/
prod(size(f));
PSNR = 10*log(1/MSE1)/log(10)
MSE11 = sum(sum((double(f)/255-output_image1).^2))/
prod(size(f));
PSNR1 = 10*log(1/MSE11)/log(10);

end

%Original clipping + companding + AWGN ...............

%Transmitter .........................................
%Data Generation .....................................

f = imread ('Cameraman.tif');
[M,N] = size(f) ;
g1 = im2col(f, [M,N], [M,N], 'distinct');
h1 = dec2bin(double(g1));
[M1,N1] = size(h1) ;
z1 = zeros (M1,N1) ;
for i = 1: M1
for j = 1: N1
z1(i,j) = str2num(h1(i,j));
end;
end;
[R1,T1] = size(z1) ;
zz1 = reshape(z1,R1*T1, 1);
% The transmitted data ...............................
zz = [zz1];

%ENCODING ............................................
trel = poly2trellis(7,[171 133]);% Trellis
data1 = zz;
298 App en d i x B

zz = convenc(data1,trel);
s1 = length(zz);
para = 128;
nd = 6; %number of information OFDM symbol for one loop
ml = 2; %Modulation level: QPSK
sr = 250000; %symbol rate
br = sr.*ml; %Bit rat per carrier
Ipoint = 8;%Number of over samples
gilen = 32;
flat = 1;
fd = 600;

dctlen = para;
noc = para;
ofdm_length = para*nd*ml; %Total no for one loop

%Dividing the image into blocks ......................

nloops = ceil((length(zz))/ofdm_length);
new_data = nloops*ofdm_length ;
nzeros = new_data - length(zz);
input_data = [zz;zeros(nzeros,1)];
input_data2 = reshape(input_data,ofdm_length,nloops);

%Transmission ON FFT_OFDM .............................

for ebno = [0,2,4,6,8,10];

demodata1 = zeros(ofdm_length,nloops);
for jj = 1: nloops % loop for columns
serdata1 = input_data2(:,jj)';
demodata = 
fft_channel_clipping_companding(serda
ta1,para,nd,ml,gilen,dctlen,sr,ebno,
br);
demodata1(:,jj) = 
demodata(: ); % the
output of ofdm
columns

end

%Received image .............................. .......

[Mr,Nr] = size(demodata1);
% demodata2 = demodata1(: );
yy = reshape (demodata1,Mr*Nr,1);%
part1 = yy(1: s1);
yy11 = part1;
 App en d i x B 299

%DECODING ............................................
yy11 = vitdec(yy11',trel,1,'term','hard');% Decode.

yy1 = reshape(yy11,[R1,T1]);
for i = 1: M1
for j = 1: N1
zn1(i,j) = num2str(yy1(i,j));
end;
end;
hn1 = bin2dec(zn1);
gn1 = col2im(hn1, [M,N], [M,N], 'distinct');
%The Error between Trans ..............................
output_image = gn1/255;

output_image1 = medfilt2(output_image);

%The Error between Trans ..............................

MSE1 = sum(sum((double(f)/255-output_image).^2))/
prod(size(f));
PSNR = 10*log(1/MSE1)/log(10)

MSE11 = sum(sum((double(f)/255-output_image1).^2))/
prod(size(f));
PSNR1 = 10*log(1/MSE11)/log(10);

end

%Transmission ON DCT_OFDM .............................

for ebno = [0,2,4,6,8,10];

demodata1 = zeros(ofdm_length,nloops);
for jj = 1: nloops % loop for columns
serdata1 = input_data2(:,jj)';
demodata = 
dct_channel_clipping_companding(serda
ta1,para,nd,ml,gilen,dctlen,sr,ebno,
br);
 demodata1(:,jj) = demodata(: ); % the
output of ofdm columns

end

%Received image ......................................

[Mr,Nr] = size(demodata1);
% demodata2 = demodata1(: );
300 App en d i x B

yy = reshape (demodata1,Mr*Nr,1);%
part1 = yy(1: s1);
yy11 = part1;
%DECODING ............................................
yy11 = vitdec(yy11',trel,1,'term','hard');% Decode.

yy1 = reshape(yy11,[R1,T1]);
for i = 1: M1
for j = 1: N1
zn1(i,j) = num2str(yy1(i,j));
end;
end;
hn1 = bin2dec(zn1);
gn1 = col2im(hn1, [M,N], [M,N], 'distinct');
%The Error between Trans .......................
output_image = gn1/255;
output_image1 = medfilt2(output_image);

%The Error between Trans ..............................

MSE1 = s
um(sum((double(f)/255-output_image).^2))/
prod(size(f));
PSNR = 10*log(1/MSE1)/log(10)

MSE11 = s
um(sum((double(f)/255-output_image1).^2))/
prod(size(f));
PSNR1 = 10*log(1/MSE11)/log(10);

end

%Transmission ON DWT_OFDM .............................

for ebno = [0,2,4,6,8,10];

demodata1 = zeros(ofdm_length,nloops);
for jj = 1: nloops % loop for columns
serdata1 = input_data2(:,jj)';
demodata = 
dwt_channel_clipping_companding(serda
ta1,para,nd,ml,gilen,dctlen,sr,ebno,
br);
demodata1(:,jj) = 
demodata(: ); % the
output of ofdm
columns

end
 App en d i x B 3 01

%Received image ......................................

[Mr,Nr] = size(demodata1);
% demodata2 = demodata1(: );
yy = reshape (demodata1,Mr*Nr,1);%
part1 = yy(1: s1);
yy11 = part1;
%DECODING ............................................
yy11 = vitdec(yy11',trel,1,'term','hard');% Decode.

yy1 = reshape(yy11,[R1,T1]);
for i = 1: M1
for j = 1: N1
zn1(i,j) = num2str(yy1(i,j));
end;
end;
hn1 = bin2dec(zn1);
gn1 = col2im(hn1, [M,N], [M,N], 'distinct');
%The Error between Trans ..............................
output_image = gn1/255;

output_image1 = medfilt2(output_image);

%The Error between Trans ..............................

MSE1 = s
um(sum((double(f)/255-output_image).^2))/
prod(size(f));
PSNR = 10*log(1/MSE1)/log(10)

MSE11 = s
um(sum((double(f)/255-output_image1).^2))/
prod(size(f));
PSNR1 = 10*log(1/MSE11)/log(10);

end

%Image without Encryption + clipping over

estimated channel ............... ...........
­
%Transmitter .........................................
%Data Generation .....................................
f = imread ('Cameraman.tif');
[M,N] = size(f) ;
g1 = im2col(f, [M,N], [M,N], 'distinct');
h1 = dec2bin(double(g1));
[M1,N1] = size(h1) ;
z1 = zeros (M1,N1) ;
for i = 1: M1
for j = 1: N1
302 App en d i x B

z1(i,j) = str2num(h1(i,j));
end;
end;
[R1,T1] = size(z1) ;
zz1 = reshape(z1,R1*T1, 1);
% The transmitted data ............... ...............
zz = [zz1];

%ENCODING ............................................
trel = poly2trellis(7,[171 133]);% Trellis
data1 = zz;
zz = convenc(data1,trel);
s1 = length(zz);
para = 128;
nd = 6; 
%number of information OFDM symbol for one
loop
m1 = 2; %Modulation level: QPSK
sr = 250000; %symbol rate
br = sr.*m1; %Bit rat per carrier
Ipoint = 8;%Number of over samples
gilen = 32;
flat = 1;
fd = 600;
fftlen = para;
noc = para;
ofdm_length = para*nd*m1; %Total no for one
loop

%Dividing the image into blocks ......................

nloops = ceil((length(zz))/ofdm_length);
new_data = nloops*ofdm_length ;
nzeros = new_data - length(zz);
input_data = [zz;zeros(nzeros,1)];
input_data2 = reshape(input_data,ofdm_length,nloops);

%Transmission ON FFT_OFDM .............................

for ebno = [0,2,4,6,8,10];

demodata1 = zeros(ofdm_length,nloops);
for jj = 1: nloops % loop for columns
serdata = input_data2(:,jj)';
demodata = 
fft_channel_estimation_no_mapping_cli
pping(serdata,para,nd,m1,gilen,fftlen
,sr,ebno, br,fd,flat);
 App en d i x B 303

demodata1(:,jj) = 
demodata(: ); % the
output of ofdm
columns

end

%Received image ......................................

[Mr,Nr] = size(demodata1);
% demodata2 = demodata1(: );
yy = reshape (demodata1,Mr*Nr,1);%
part1 = yy(1: s1);
yy11 = part1;
%DECODING ............................................
yy11 = vitdec(yy11',trel,1,'term','hard');% Decode.

yy1 = reshape(yy11,[R1,T1]);
for i = 1: M1
for j = 1: N1
zn1(i,j) = num2str(yy1(i,j));
end;
end;
hn1 = bin2dec(zn1);
gn1 = col2im(hn1, [M,N], [M,N], 'distinct');
%The Error between Trans .......................
output_image = gn1/255;

output_image1 = medfilt2(output_image);

%The Error between Trans ..............................

MSE1 = s
um(sum((double(f)/255-output_image).^2))/
prod(size(f));
PSNR = 10*log(1/MSE1)/log(10)

MSE11 = s
um(sum((double(f)/255-output_image1).^2))/
prod(size(f));
PSNR1 = 10*log(1/MSE11)/log(10);

end

%Transmission ON DCT_OFDM ......................

for ebno = [0,2,4,6,8,10];

demodata1 = zeros(ofdm_length,nloops);
for jj = 1: nloops % loop for columns
304 App en d i x B

serdata = input_data2(:,jj)';
demodata = 
dct_channel_estimation_no_mapping_cli
pping(serdata,para,nd,m1,gilen,fftlen
,sr,ebno, br,fd,flat);
demodata1(:,jj) = 
demodata(: ); % the
output of ofdm
columns

end

%Received image ......................................

[Mr,Nr] = size(demodata1);
% demodata2 = demodata1(: );
yy = reshape (demodata1,Mr*Nr,1);%
part1 = yy(1: s1);
yy11 = part1;
%DECODING ............................................
yy11 = vitdec(yy11',trel,1,'term','hard');% Decode.

yy1 = reshape(yy11,[R1,T1]);
for i = 1: M1
for j = 1: N1
zn1(i,j) = num2str(yy1(i,j));
end;
end;
hn1 = bin2dec(zn1);
gn1 = col2im(hn1, [M,N], [M,N], 'distinct');
%The Error between Trans ............................
output_image = gn1/255;

output_image1 = medfilt2(output_image);

%The Error between Trans ..............................

MSE1 = s
um(sum((double(f)/255-output_image).^2))/
prod(size(f));
PSNR = 10*log(1/MSE1)/log(10)

MSE11 = s
um(sum((double(f)/255-output_image1).^2))/
prod(size(f));
PSNR1 = 10*log(1/MSE11)/log(10);

end

%Transmission ON DWT_OFDM .........................

 App en d i x B 305

for ebno = [0,2,4,6,8,10];

demodata1 = zeros(ofdm_length,nloops);
for jj = 1: nloops % loop for columns
serdata = input_data2(:,jj)';
demodata = 
dwt_channel_estimation_no_mapping_cli
pping(serdata,para,nd,m1,gilen,fftlen
,sr,ebno, br,fd,flat);
demodata1(:,jj) = demodata(: ); % the
output of ofdm
columns

end

%Received image ......................................

[Mr,Nr] = size(demodata1);
% demodata2 = demodata1(: );
yy = reshape (demodata1,Mr*Nr,1);%
part1 = yy(1: s1);
yy11 = part1;
%DECODING ............................................
yy11 = vitdec(yy11',trel,1,'term','hard');% Decode.

yy1 = reshape(yy11,[R1,T1]);
for i = 1: M1
for j = 1: N1
zn1(i,j) = num2str(yy1(i,j));
end;
end;
hn1 = bin2dec(zn1);
gn1 = col2im(hn1, [M,N], [M,N], 'distinct');
%The Error between Trans ............................
output_image = gn1/255;

output_image1 = medfilt2(output_image);

%The Error between Trans ..............................

MSE1 = s
um(sum((double(f)/255-output_image).^2))/
prod(size(f));
PSNR = 10*log(1/MSE1)/log(10)

MSE11 = s
um(sum((double(f)/255-output_image1).^2))/
prod(size(f));
PSNR1 = 10*log(1/MSE11)/log(10);

end
306 App en d i x B

%Image with no Encryption over OFDM system +

offset + cyclic prefix + AWGN ...............
%Transmitter .........................................
%Data Generation .....................................
%Transmitter ............... ..........................

%Data Generation .....................................

f = imread ('Cameraman.tif');
[M,N] = size(f) ;
g1 = im2col(f, [M,N], [M,N], 'distinct');
h1 = dec2bin(double(g1));
[M1,N1] = size(h1) ;
z1 = zeros (M1,N1) ;
for i = 1: M1
for j = 1: N1
z1(i,j) = str2num(h1(i,j));
end;
end;
[R1,T1] = size(z1) ;
zz1 = reshape(z1,R1*T1, 1);
% The transmitted data ...............................
zz = [zz1];
trel = poly2trellis(7,[171 133]);% Trellis
data1 = zz;
zz = convenc(data1,trel);
s1 = length(zz);
para = 128;
nd = 6; %number of information OFDM symbol for one loop
m1 = 2; %Modulation level: QPSK
sr = 250000; %symbol rate
br = sr.*ml; %Bit rate per carrier
Ipoint = 8;%Number of over samples
gilen = 32;
flat = 1;
fd = 600;
epsilon = 0.1;

fftlen = para;
noc = para;
ofdm_length = para*nd*ml; %Total no for one loop

%Dividing the image into blocks ......................

nloops = ceil((length(zz))/ofdm_length);
new_data = nloops*ofdm_length ;
 App en d i x B 307

nzeros = new_data - length(zz);

input_data = [zz;zeros(nzeros,1)];
input_data2 = reshape(input_data,ofdm_length,nloops);

%Transmission ON FFT_OFDM .............................

for ebno = [0,2,4,6,8,10];

demodata1 = zeros(ofdm_length,nloops);
for jj = 1: nloops % loop for columns
serdata = input_data2(:,jj)';
demodata = 
fft_channel_offset_g(serdata,para,nd,
m1,gilen,fftlen,sr,ebno, br,epsilon);
demodata1(:,jj) = demodata(: ); % the
output of ofdm
columns

end

%Received image ......................................

[Mr,Nr] = size(demodata1);
% demodata2 = demodata1(: );
yy = reshape (demodata1,Mr*Nr,1);%
part1 = yy(1: s1);
yy11 = part1;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%DECODING ...........
yy11 = vitdec(yy11',trel,1,'term','hard');% Decode.

yy1 = reshape(yy11,[R1,T1]);
for i = 1: M1
for j = 1: N1
zn1(i,j) = num2str(yy1(i,j));
end;
end;
hn1 = bin2dec(zn1);
gn1 = col2im(hn1, [M,N], [M,N], 'distinct');
%The Error between Trans ............................
output_image = gn1/255;

output_image1 = medfilt2(output_image);

%The Error between Trans ..............................

MSE1 = s
um(sum((double(f)/255-output_image).^2))/
prod(size(f));
PSNR = 10*log(1/MSE1)/log(10)
308 App en d i x B

MSE11 = s
um(sum((double(f)/255-output_image1).^2))/
prod(size(f));
PSNR1 = 10*log(1/MSE11)/log(10);

end

%Transmission ON DCT_OFDM ......................

for ebno = [0,2,4,6,8,10];

demodata1 = zeros(ofdm_length,nloops);
for jj = 1: nloops % loop for columns
serdata1 = input_data2(:,jj)';
demodata = 
dct_channel_offset_g(serdata1,para,​
nd,m1,gilen,fftlen,sr,ebno,
br,epsilon);
demodata1(:,jj) = 
demodata(: ); % the
output of ofdm
columns

end

%Received image ......................................

[Mr,Nr] = size(demodata1);
% demodata2 = demodata1(: );
yy = reshape (demodata1,Mr*Nr,1);%
part1 = yy(1: s1);
yy11 = part1;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%DECODING ...........
yy11 = vitdec(yy11',trel,1,'term','hard');% Decode.

yy1 = reshape(yy11,[R1,T1]);
for i = 1: M1
for j = 1: N1
zn1(i,j) = num2str(yy1(i,j));
end;
end;
hn1 = bin2dec(zn1);
gn1 = col2im(hn1, [M,N], [M,N], 'distinct');
%The Error between Trans ............................
output_image = gn1/255;

output_image1 = medfilt2(output_image);

%The Error between Trans ..............................

 App en d i x B 309

MSE1 = s
um(sum((double(f)/255-output_image).^2))/
prod(size(f));
PSNR = 10*log(1/MSE1)/log(10)

MSE11 = s
um(sum((double(f)/255-output_image1).^2))/
prod(size(f));
PSNR1 = 10*log(1/MSE11)/log(10);

end

%Transmission ON DWT_OFDM ........................

for ebno = [0,2,4,6,8,10];

demodata1 = zeros(ofdm_length,nloops);
for jj = 1: nloops % loop for columns
serdata1 = input_data2(:,jj)';
demodata = 
dwt_channel_offset_g(serdata1,para,​
nd,m1,gilen,fftlen,sr,ebno,
br,epsilon);
demodata1(:,jj) = 
demodata(: ); % the
output of ofdm
columns

end

%Received image ......................................

[Mr,Nr] = size(demodata1);
% demodata2 = demodata1(: );
yy = reshape (demodata1,Mr*Nr,1);%
part1 = yy(1: s1);
yy11 = part1;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%DECODING ...........
yy11 = vitdec(yy11',trel,1,'term','hard');% Decode.

yy1 = reshape(yy11,[R1,T1]);
for i = 1: M1
for j = 1: N1
zn1(i,j) = num2str(yy1(i,j));
end;
end;
hn1 = bin2dec(zn1);
gn1 = col2im(hn1, [M,N], [M,N], 'distinct');
%The Error between Trans ............................
output_image = gn1/255;
310 App en d i x B

output_image1 = medfilt2(output_image);

%The Error between Trans ..............................

MSE1 = s
um(sum((double(f)/255-output_image).^2))/
prod(size(f));
PSNR = 10*log(1/MSE1)/log(10)

MSE11 = s
um(sum((double(f)/255-output_image1).^2))/
prod(size(f));
PSNR1 = 10*log(1/MSE11)/log(10);

end

%Image with no Encryption over OFDM system + offset +

zero padding + AWGN ...............
%Transmitter .........................................
%Data Generation .....................................

f = imread ('Cameraman.tif');
[M,N] = size(f) ;
g1 = im2col(f, [M,N], [M,N], 'distinct');
h1 = dec2bin(double(g1));
[M1,N1] = size(h1) ;
z1 = zeros (M1,N1) ;
for i = 1: M1
for j = 1: N1
z1(i,j) = str2num(h1(i,j));
end;
end;
[R1,T1] = size(z1) ;
zz1 = reshape(z1,R1*T1, 1);
% The transmitted data ............... ...............
............... .......
zz = [zz1];
trel = poly2trellis(7,[171 133]);% Trellis
data1 = zz;
zz = convenc(data1,trel);
s1 = length(zz);
para = 128;
nd = 6; %number of information OFDM symbol for one loop
m1 = 2; %Modulation level: QPSK
sr = 250000; %symbol rate
br = sr.*ml; %Bit rate per carrier
Ipoint = 8;%Number of over samples
gilen = 32;
 App en d i x B 311

flat = 1;
fd = 600;
epsilon = 0.1;
fftlen = para;
noc = para;
ofdm_length = para*nd*ml; %Total no for one loop

%Dividing the image into blocks ......................

nloops = ceil((length(zz))/ofdm_length);
new_data = nloops*ofdm_length ;
nzeros = new_data - length(zz);
input_data = [zz;zeros(nzeros,1)];
input_data2 = reshape(input_data,ofdm_length,nloops);

%Transmission ON FFT_OFDM .............................

for ebno = [0,2,4,6,8,10];

demodata1 = zeros(ofdm_length,nloops);
for jj = 1: nloops % loop for columns
serdata = input_data2(:,jj)';
demodata = 
fft_channel_offset_zg(serdata,para,​
nd,m1,gilen,fftlen,sr,ebno,
br,epsilon);
demodata1(:,jj) = demodata(: ); % the
output of ofdm
columns

end

%Received image ......................................

[Mr,Nr] = size(demodata1);
% demodata2 = demodata1(: );
yy = reshape (demodata1,Mr*Nr,1);%
part1 = yy(1: s1);
yy11 = part1;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%DECODING ...........
yy11 = vitdec(yy11',trel,1,'term','hard');% Decode.

yy1 = reshape(yy11,[R1,T1]);
for i = 1: M1
for j = 1: N1
zn1(i,j) = num2str(yy1(i,j));
end;
end;
312 App en d i x B

hn1 = bin2dec(zn1);
gn1 = col2im(hn1, [M,N], [M,N], 'distinct');
%The Error between Trans ............................
output_image = gn1/255;

output_image1 = medfilt2(output_image);

%The Error between Trans ..............................

MSE1 = s
um(sum((double(f)/255-output_image).^2))/
prod(size(f));
PSNR = 10*log(1/MSE1)/log(10)

MSE11 = s
um(sum((double(f)/255-output_image1).^2))/
prod(size(f));
PSNR1 = 10*log(1/MSE11)/log(10);

end

%Transmission ON DCT_OFDM .............................

for ebno = [0,2,4,6,8,10];

demodata1 = zeros(ofdm_length,nloops);
for jj = 1: nloops % loop for columns
serdata1 = input_data2(:,jj)';
demodata = 
dct_channel_offset_zg(serdata1,para,n
d,m1,gilen,fftlen,sr,ebno,
br,epsilon);
demodata1(:,jj) = 
demodata(: ); % the
output of ofdm
columns

end

%Received image ......................................

[Mr,Nr] = size(demodata1);
% demodata2 = demodata1(: );
yy = reshape (demodata1,Mr*Nr,1);%
part1 = yy(1: s1);
yy11 = part1;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%DECODING ...........
yy11 = vitdec(yy11',trel,1,'term','hard');% Decode.

yy1 = reshape(yy11,[R1,T1]);
 App en d i x B 313

for i = 1: M1
for j = 1: N1
zn1(i,j) = num2str(yy1(i,j));
end;
end;
hn1 = bin2dec(zn1);
gn1 = col2im(hn1, [M,N], [M,N], 'distinct');
%The Error between Trans ............................
output_image = gn1/255;

output_image1 = medfilt2(output_image);

%The Error between Trans ..............................

MSE1 = s
um(sum((double(f)/255-output_image).^2))/
prod(size(f));
PSNR = 10*log(1/MSE1)/log(10)

MSE11 = s
um(sum((double(f)/255-output_image1).^2))/
prod(size(f));
PSNR1 = 10*log(1/MSE11)/log(10);

end

%Transmission ON DWT_OFDM .............................

for ebno = [0,2,4,6,8,10];

demodata1 = zeros(ofdm_length,nloops);
for jj = 1: nloops % loop for columns
serdata1 = input_data2(:,jj)';
demodata = 
dwt_channel_offset_zg(serdata1,para,n
d,m1,gilen,fftlen,sr,ebno,
br,epsilon);
demodata1(:,jj) = 
demodata(: ); % the
output of ofdm
columns

end

%Received image ......................................

[Mr,Nr] = size(demodata1);
% demodata2 = demodata1(: );
yy = reshape (demodata1,Mr*Nr,1);%
part1 = yy(1: s1);
yy11 = part1;
314 App en d i x B

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%DECODING ...........
yy11 = vitdec(yy11',trel,1,'term','hard');% Decode.

yy1 = reshape(yy11,[R1,T1]);
for i = 1: M1
for j = 1: N1
zn1(i,j) = num2str(yy1(i,j));
end;
end;
hn1 = bin2dec(zn1);
gn1 = col2im(hn1, [M,N], [M,N], 'distinct');
%The Error between Trans ............................
output_image = gn1/255;

output_image1 = medfilt2(output_image);

%The Error between Trans ..............................

MSE1 = s
um(sum((double(f)/255-output_image).^2))/
prod(size(f));
PSNR = 10*log(1/MSE1)/log(10)

MSE11 = s
um(sum((double(f)/255-output_image1).^2))/
prod(size(f));
PSNR1 = 10*log(1/MSE11)/log(10);

end

%Image with no Encryption + OFDM + offset + cyclic

prefix + over estimated channel...
%Transmitter .........................................
%Data Generation .....................................

%Transmitter .........................................

%Data Generation .....................................

f = imread ('Cameraman.tif');
[M,N] = size(f) ;
g1 = im2col(f, [M,N], [M,N], 'distinct');
h1 = dec2bin(double(g1));
[M1,N1] = size(h1) ;
z1 = zeros (M1,N1) ;
for i = 1: M1
for j = 1: N1
z1(i,j) = str2num(h1(i,j));
end;
 App en d i x B 315

end;
[R1,T1] = size(z1) ;
zz1 = reshape(z1,R1*T1, 1);
% The transmitted data ...............................
zz = [zz1];

%ENCODING ............................................
trel = poly2trellis(7,[171 133]);% Trellis
data1 = zz;
zz = convenc(data1,trel);
s1 = length(zz);
para = 128;
nd = 6; %
number of information OFDM symbol for one
loop
m1 = 2; %Modulation level: QPSK
sr = 250000; %symbol rate
br = sr.*m1; %Bit rate per carrier
Ipoint = 8;%Number of over samples
gilen = 32;
flat = 1;
fd = 600;

fftlen = para;
noc = para;
ofdm_length = para*nd*m1; %Total no for one loop

%Dividing the image into blocks ......................

nloops = ceil((length(zz))/ofdm_length);
new_data = nloops*ofdm_length ;
nzeros = new_data - length(zz);
input_data = [zz;zeros(nzeros,1)];
input_data2 = reshape(input_data,ofdm_length,nloops);

%Transmission ON FFT_OFDM .............................

for ebno = [0,2,4,6,8,10];

demodata1 = zeros(ofdm_length,nloops);
for jj = 1: nloops % loop for columns
serdata = input_data2(:,jj)';
demodata = 
fft_channel_estimation_no_mapping_g(s
erdata,para,nd,m1,gilen,fftlen,sr,e
bno, br,fd,flat);
316 App en d i x B

demodata1(:,jj) = 
demodata(: ); % the
output of ofdm
columns

end

%Received image ......................................

[Mr,Nr] = size(demodata1);
% demodata2 = demodata1(: );
yy = reshape (demodata1,Mr*Nr,1);%
part1 = yy(1: s1);
yy11 = part1;
%DECODING ............................................
yy11 = vitdec(yy11',trel,1,'term','hard');% Decode.

yy1 = reshape(yy11,[R1,T1]);
for i = 1: M1
for j = 1: N1
zn1(i,j) = num2str(yy1(i,j));
end;
end;
hn1 = bin2dec(zn1);
gn1 = col2im(hn1, [M,N], [M,N], 'distinct');
%The Error between Trans ............................
output_image = gn1/255;

output_image1 = medfilt2(output_image);

%The Error between Trans ............................

MSE1 = s
um(sum((double(f)/255-output_image).^2))/
prod(size(f));
PSNR = 10*log(1/MSE1)/log(10)

MSE11 = s
um(sum((double(f)/255-output_image1).^2))/
prod(size(f));
PSNR1 = 10*log(1/MSE11)/log(10);

end

%Transmission ON DCT_OFDM .............................

for ebno = [0,2,4,6,8,10];

demodata1 = zeros(ofdm_length,nloops);
for jj = 1: nloops % loop for columns
 App en d i x B 317

serdata = input_data2(:,jj)';
demodata = 
dct_channel_estimation_no_mapping_g(s
erdata,para,nd,m1,gilen,fftlen,sr,e
bno, br,fd,flat);
demodata1(:,jj) = demodata(: ); % the
output of ofdm
columns

end

%Received image ......................................

[Mr,Nr] = size(demodata1);
% demodata2 = demodata1(: );
yy = reshape (demodata1,Mr*Nr,1);%
part1 = yy(1: s1);
yy11 = part1;
%DECODING ............................................
yy11 = vitdec(yy11',trel,1,'term','hard');% Decode.

yy1 = reshape(yy11,[R1,T1]);
for i = 1: M1
for j = 1: N1
zn1(i,j) = num2str(yy1(i,j));
end;
end;
hn1 = bin2dec(zn1);
gn1 = col2im(hn1, [M,N], [M,N], 'distinct');
%The Error between Trans ............................
output_image = gn1/255;

output_image1 = medfilt2(output_image);

%The Error between Trans ..............................

MSE1 = s
um(sum((double(f)/255-output_image).^2))/
prod(size(f));
PSNR = 10*log(1/MSE1)/log(10)

MSE11 = s
um(sum((double(f)/255-output_image1).^2))/
prod(size(f));
PSNR1 = 10*log(1/MSE11)/log(10);

end

%Transmission ON DWT_OFDM ......................

318 App en d i x B

for ebno = [0,2,4,6,8,10];

demodata1 = zeros(ofdm_length,nloops);
for jj = 1: nloops % loop for columns
serdata = input_data2(:,jj)';
demodata = 
dwt_channel_estimation_no_mapping_g(s
erdata,para,nd,m1,gilen,fftlen,sr,e
bno, br,fd,flat);
demodata1(:,jj) = demodata(: ); % the
output of ofdm
columns

end

%Received image ......................................

[Mr,Nr] = size(demodata1);
% demodata2 = demodata1(: );
yy = reshape (demodata1,Mr*Nr,1);%
part1 = yy(1: s1);
yy11 = part1;
%DECODING ............... .............................
yy11 = vitdec(yy11',trel,1,'term','hard');% Decode.

yy1 = reshape(yy11,[R1,T1]);
for i = 1: M1
for j = 1: N1
zn1(i,j) = num2str(yy1(i,j));
end;
end;
hn1 = bin2dec(zn1);
gn1 = col2im(hn1, [M,N], [M,N], 'distinct');
%The Error between Trans ............................
output_image = gn1/255;

output_image1 = medfilt2(output_image);

%The Error between Trans ..............................

MSE1 = s
um(sum((double(f)/255-output_image).^2))/
prod(size(f));
PSNR = 10*log(1/MSE1)/log(10)

MSE11 = s
um(sum((double(f)/255-output_image1).^2))/
prod(size(f));
PSNR1 = 10*log(1/MSE11)/log(10);

end
 App en d i x B 319

%Image with no Encryption + OFDM + offset + zero

padding + over estimated channel........
%Transmitter .........................................
%Data Generation .....................................
%Transmitter .........................................
%Data Generation .....................................
f = imread ('Cameraman.tif');
[M,N] = size(f) ;
g1 = im2col(f, [M,N], [M,N], 'distinct');
h1 = dec2bin(double(g1));
[M1,N1] = size(h1) ;
z1 = zeros (M1,N1) ;
for i = 1: M1
for j = 1: N1
z1(i,j) = str2num(h1(i,j));
end;
end;
[R1,T1] = size(z1) ;
zz1 = reshape(z1,R1*T1, 1);
% The transmitted data ...............................
zz = [zz1];

%ENCODING ............................................
trel = poly2trellis(7,[171 133]);% Trellis
data1 = zz;
zz = convenc(data1,trel);
s1 = length(zz);
para = 128;
nd = 6; %number of information OFDM symbol for one
loop
m1 = 2; %Modulation level: QPSK
sr = 250000; %symbol rate
br = sr.*m1; %Bit rate per carrier
Ipoint = 8;%Number of over samples
gilen = 32;
flat = 1;
fd = 600;

fftlen = para;
noc = para;
ofdm_length = para*nd*m1; %Total no for one loop

%Dividing the image into blocks ......................

nloops = ceil((length(zz))/ofdm_length);
320 App en d i x B

new_data = nloops*ofdm_length ;
nzeros = new_data - length(zz);
input_data = [zz;zeros(nzeros,1)];
input_data2 = reshape(input_data,ofdm_length,nloops);

%Transmission ON FFT_OFDM .............................

for ebno = [0,2,4,6,8,10];

demodata1 = zeros(ofdm_length,nloops);
for jj = 1: nloops % loop for columns
serdata = input_data2(:,jj)';
demodata = 
fft_channel_estimation_no_mapping_zg​
(serdata,para,nd,m1,gilen,fftlen,sr,e
bno, br,fd,flat);
demodata1(:,jj) = 
demodata(: ); % the
output of ofdm
columns

end

%Received image ......................................

[Mr,Nr] = size(demodata1);
% demodata2 = demodata1(: );
yy = reshape (demodata1,Mr*Nr,1);%
part1 = yy(1: s1);
yy11 = part1;
%DECODING ............................................
yy11 = vitdec(yy11',trel,1,'term','hard');% Decode.

yy1 = reshape(yy11,[R1,T1]);
for i = 1: M1
for j = 1: N1
zn1(i,j) = num2str(yy1(i,j));
end;
end;
hn1 = bin2dec(zn1);
gn1 = col2im(hn1, [M,N], [M,N], 'distinct');
%The Error between Trans ............................
output_image = gn1/255;

output_image1 = medfilt2(output_image);

%The Error between Trans ..............................

 App en d i x B 3 21

MSE1 = s
um(sum((double(f)/255-output_image).^2))/
prod(size(f));
PSNR = 10*log(1/MSE1)/log(10)

MSE11 = s
um(sum((double(f)/255-output_image1).^2))/
prod(size(f));
PSNR1 = 10*log(1/MSE11)/log(10);

end

%Transmission ON DCT_OFDM .........................

for ebno = [0,2,4,6,8,10];

demodata1 = zeros(ofdm_length,nloops);
for jj = 1: nloops % loop for columns
serdata = input_data2(:,jj)';
demodata = 
dct_channel_estimation_no_mapping_zg​
(serdata,para,nd,m1,gilen,fftlen,sr,e
bno, br,fd,flat);
demodata1(:,jj) = demodata(: ); % the
output of ofdm
columns

end

%Received image ......................................

[Mr,Nr] = size(demodata1);
% demodata2 = demodata1(: );
yy = reshape (demodata1,Mr*Nr,1);%
part1 = yy(1: s1);
yy11 = part1;
%DECODING ............................................
yy11 = vitdec(yy11',trel,1,'term','hard');% Decode.

yy1 = reshape(yy11,[R1,T1]);
for i = 1: M1
for j = 1: N1
zn1(i,j) = num2str(yy1(i,j));
end;
end;
hn1 = bin2dec(zn1);
gn1 = col2im(hn1, [M,N], [M,N], 'distinct');
%The Error between Trans ............................
322 App en d i x B

output_image = gn1/255;

output_image1 = medfilt2(output_image);

%The Error between Trans ..............................

MSE1 = s
um(sum((double(f)/255-output_image).^2))/
prod(size(f));
PSNR = 10*log(1/MSE1)/log(10)

MSE11 = s
um(sum((double(f)/255-output_image1).^2))/
prod(size(f));
PSNR1 = 10*log(1/MSE11)/log(10);

end

%Transmission ON DWT_OFDM ..........................

for ebno = [0,2,4,6,8,10];

demodata1 = zeros(ofdm_length,nloops);
for jj = 1: nloops % loop for columns
serdata = input_data2(:,jj)';
demodata = 
dwt_channel_estimation_no_mapping_zg(
serdata,para,nd,m1,gilen,fftlen,sr,e
bno, br,fd,flat);
demodata1(:,jj) = 
demodata(: ); % the
output of ofdm
columns

end

%Received image ......................................

[Mr,Nr] = size(demodata1);
% demodata2 = demodata1(: );
yy = reshape (demodata1,Mr*Nr,1);%
part1 = yy(1: s1);
yy11 = part1;
%DECODING ............................................
yy11 = vitdec(yy11',trel,1,'term','hard');% Decode.

yy1 = reshape(yy11,[R1,T1]);
for i = 1: M1
for j = 1: N1
zn1(i,j) = num2str(yy1(i,j));
end;
 App en d i x B 323

end;
hn1 = bin2dec(zn1);
gn1 = col2im(hn1, [M,N], [M,N], 'distinct');
%The Error between Trans ............................
output_image = gn1/255;

output_image1 = medfilt2(output_image);

%The Error between Trans ..............................

MSE1 = s
um(sum((double(f)/255-output_image).^2))/
prod(size(f));
PSNR = 10*log(1/MSE1)/log(10)

MSE11 = s
um(sum((double(f)/255-output_image1).^2))/
prod(size(f));
PSNR1 = 10*log(1/MSE11)/log(10);

end

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function outdemodata = 
fft_channel(serdata,para,nd,m1,
gilen,fftlen,sr,ebno, br)

%serial to parallel conversion .......................

paradata = reshape(serdata,para,nd*m1);
%QPSK modulation .....................................
[ich,qch] = qpskmod(paradata,para,nd,m1);
% [ich0,qch0] = compoversamp(ich01,qch01,length(ich01)
,Ipoint);
kmod = 1/sqrt(2);
ich1 = ich.*kmod;
qch1 = qch.*kmod;
%IFFT ............... ................................
x = ich1 + qch1.*j;
y = ifft(x);
ich2 = real (y);
qch2 = imag (y);

%Guard interval insertion .............................

[ich3,qch3] = giins(ich2,qch2,fftlen,gilen,nd);
fftlen2 = fftlen + gilen;

%Attenuation Calculation ..............................

324 App en d i x B

spow = sum(ich3.^2+qch3.^2)/nd./para;
attn = 0.5*spow*sr/br*10.^(-ebno/10);
attn = sqrt (attn);

%Receiver ...........................................
%AWGN addition .......................................

[ich4,qch4] = comb(ich3,qch3,attn);

%Guard interval removal ...............................

[ich5,qch5] = girem (ich4,qch4,fftlen2,gilen,nd);
%FFT ............... .................................
rx = ich5 + qch5.*j;
ry = fft(rx);
ich6 = real (ry);
qch6 = imag (ry);

%Demodulation ........................................
ich7 = ich6./kmod;
qch7 = qch6./kmod;
outdemodata = qpskdemod (ich7,qch7,para,nd,m1);
function outdemodata = 
dct_channel(serdata,para,nd,m1,
gilen,dctlen,sr,ebno, br)

%serial to parallel conversion .......................

paradata = reshape(serdata,para,nd*m1);
%QPSK modulation .....................................
[ich,qch] = qpskmod(paradata,para,nd,m1);
% [ich0,qch0] =
compoversamp(ich01,qch01,length(ich01),Ipoint);
kmod = 1/sqrt(2);
ich1 = ich.*kmod;
qch1 = qch.*kmod;
%IDCT ................................................
x = ich1 + qch1.*j;
y = idct(x);
ich2 = real (y);
qch2 = imag (y);

%Guard interval insertion .............................

[ich3,qch3] = giins(ich2,qch2,dctlen,gilen,nd);
dctlen2 = dctlen + gilen;
 App en d i x B 325

%Attenuation Calculation ..............................

spow = sum(ich3.^2+qch3.^2)/nd./para;
attn = 0.5*spow*sr/br*10.^(-ebno/10);
attn = sqrt (attn);

%Receiver ...........................................

%AWGN addition .......................................

[ich4,qch4] = comb(ich3,qch3,attn);

%Guard interval removal ...............................

[ich5,qch5] = girem (ich4,qch4,dctlen2,gilen,nd);
%DCT .................................................
rx = ich5 + qch5.*j;
ry = dct(rx);
ich6 = real (ry);
qch6 = imag (ry);

%Demodulation ........................................
ich7 = ich6./kmod;
qch7 = qch6./kmod;
outdemodata = qpskdemod (ich7,qch7,para,nd,m1);

function outdemodata = 
dwt_channel(serdata,para,nd,m1,
gilen,dctlen,sr,ebno, br)

%serial to parallel conversion .......................

paradata = reshape(serdata,para,nd*m1);
%QPSK modulation .....................................
[ich,qch] = qpskmod(paradata,para,nd,m1);
% [ich0,qch0] = 
compoversamp(ich01,qch01,length​
(ich01),​Ipoint);
kmod = 1/sqrt(2);
ich1 = ich.*kmod;
qch1 = qch.*kmod;
%IDWT ................................................
x = ich1 + qch1.*j;
y = wavelet('D6',-1,x,'zpd'); % Invert 5 stages
%; % 2D wavelet transform
% R = wavelet('2D CDF 9/7',-2,Y); % Recover X from Y%
Forward transform with 5 stages
326 App en d i x B

ich2 = real (y);

qch2 = imag (y);

%Guard interval insertion .............................

[ich3,qch3] = giins(ich2,qch2,dctlen,gilen,nd);
dctlen2 = dctlen + gilen;

%Attenuation Calculation ..............................

spow = sum(ich3.^2+qch3.^2)/nd./para;
attn = 0.5*spow*sr/br*10.^(-ebno/10);
attn = sqrt (attn);

%Receiver ...........................................

%AWGN addition .......................................

[ich4,qch4] = comb(ich3,qch3,attn);

%Guard interval removal ...............................

[ich5,qch5] = girem (ich4,qch4,dctlen2,gilen,nd);
%DWT .................................................
rx = ich5 + qch5.*j;
ry = wavelet('D6',1,rx,'zpd');; % 2D wavelet transform
% R = wavelet('2D CDF 9/7',-2,Y); % Recover X from Y;
ich6 = real (ry);
qch6 = imag (ry);

%Demodulation ........................................
ich7 = ich6./kmod;
qch7 = qch6./kmod;
outdemodata = qpskdemod (ich7,qch7,para,nd,m1);

function outdemodata = 
fft_channel_clipping_companding
(serdata,para,nd,m1,gilen,fftle
n,sr,ebno, br)

paradata = reshape(serdata,para,nd*m1);
%QPSK modulation .....................................
[ich,qch] = qpskmod(paradata,para,nd,m1);
% [ich0,qch0] = 
compoversamp(ich01,qch01,length​
(ich01),Ipoint);
kmod = 1/sqrt(2);
ich1 = ich.*kmod;
 App en d i x B 327

qch1 = qch.*kmod;

%IFFT ............... ................................

x = ich1 + qch1.*i;
y = ifft(x);
CR = 4;
clipping_threshold = (10^(CR/10))*sqrt(mean(ab
s(y).^2));
tx_signal_Ang = angle(y);

for ii = 1: length(y)
if y(ii) = = y(ii);
y(ii) = y(ii);

elseif abs(y(ii)).^2> clipping_threshold

y(ii) = 
clipping_threshold.*exp(sqrt​
(-1)*tx_signal_Ang(ii));

end
end
%companding .........................
u = 4;

tx_lfdma_max = max(abs(y(1: nd*para)));

tx_lfdma_Abs = 
abs(y(1: nd*para)); % tx data
amplitude

TxSamples_lfdma1 = 
tx_lfdma_max*((log10(1+u*(tx_
lfdma_Abs./tx_lfdma_max)))/
log10(u+1)).*sign(y(1:
nd*para));

tx_ifdma_max = max(abs(y(1: nd*para)));

tx_ifdma_Abs = 
abs(y(1: nd*para)); % tx data
amplitude
yy = 
tx_ifdma_max*((log10(1+u*(tx_ifdma_Abs./
tx_ifdma_max)))/log10(u+1)).*sign(y​
(1: nd*para));

ich3 = real (yy);

qch3 = imag (yy);

%Guard interval insertion .............................

[ich4,qch4] = giins(ich3,qch3,fftlen,gilen,nd);
328 App en d i x B

fftlen2 = fftlen + gilen;

%Attenuation Calculation ............... .............

spow = sum(ich4.^2+qch4.^2)/nd./para;
attn = 0.5*spow*sr/br*10.^(-ebno/10);
attn = sqrt (attn);
%fading ..............................................
%******* Create Rayleigh fading channel object.*******
%******************* AWGN addition *******************
%Receiver ...........................................

%AWGN addition ................... ...................

[ich5,qch5] = comb(ich4,qch4,attn);
%perfect fading compensation ..........................

%Guard interval removal ..............................

[ich6,qch6] = girem (ich5,qch5,fftlen2,gilen,nd);
%FFT .................................................
rx = ich6 + qch6.*j;

%%%%%%%%%%% Expanding
rx_lfdma_Abs = abs(rx);
r_lfdma_max = tx_lfdma_max;

RxSamples_lfdma = (r_lfdma_max/u)*(exp(log10(1+u)*2.30
2585093*rx_lfdma_Abs./r_lfdma_max)-1).*sign(rx);

rx_ifdma_Abs = abs(rx);
r_ifdma_max = tx_ifdma_max;

rxx = (r_ifdma_max/u)*(exp(log10(1+u)*2.302585093
*rx_ifdma_Abs./r_ifdma_max)-1).*sign(rx);

ry = fft(rxx);
ich7 = real (ry);
qch7 = imag (ry);

%Demodulation ........................................
ich10 = ich7./kmod;
qch10 = qch7./kmod;
outdemodata = qpskdemod (ich10,qch10,para,nd,m1);

function outdemodata = dct_channel_clipping_companding

(serdata,para,nd,m1,gilen,dctlen,sr,ebno, br)
 App en d i x B 329

%serial to parallel conversion .......................

paradata = reshape(serdata,para,nd*m1);
%QPSK modulation .....................................
[ich,qch] = qpskmod(paradata,para,nd,m1);
% [ich0,qch0] = 
compoversamp(ich01,qch01,length​
(ich01),Ipoint);
kmod = 1/sqrt(2);
ich1 = ich.*kmod;
qch1 = qch.*kmod;
%IDCT ................................................
x = ich1 + qch1.*j;
y = idct(x);

CR = 4;
clipping_threshold = 
(10^(CR/10))*sqrt(mean(ab
s(y).^2));
tx_signal_Ang = angle(y);

for ii = 1: length(y)
if y(ii) = = y(ii);
y(ii) = y(ii);

elseif abs(y(ii)).^2> clipping_threshold

y(ii) = 
clipping_threshold.*exp(sqrt(-1)*tx_
signal_Ang(ii));

end
end
%companding .................................
u = 4;

tx_lfdma_max = max(abs(y(1: nd*para)));

tx_lfdma_Abs = 
abs(y(1: nd*para)); % tx data
amplitude

TxSamples_lfdma1 = 
tx_lfdma_max*((log10(1+u*​
(tx_lfdma_Abs./tx_lfdma_
max)))/log10(u+1)).*sign(y​
(1: nd*para));

tx_ifdma_max = max(abs(y(1: nd*para)));

tx_ifdma_Abs = 
abs(y(1: nd*para)); % tx data
amplitude
330 App en d i x B

yy = t
x_ifdma_max*((log10(1+u*(tx_ifdma_Abs./
tx_ifdma_max)))/log10(u+1)).*sign(y(1:
nd*para));

ich2 = real (yy);

qch2 = imag (yy);

%Guard interval insertion .............................

[ich3,qch3] = giins(ich2,qch2,dctlen,gilen,nd);
dctlen2 = dctlen + gilen;

%Attenuation Calculation .............................

spow = sum(ich3.^2+qch3.^2)/nd./para;
attn = 0.5*spow*sr/br*10.^(-ebno/10);
attn = sqrt (attn);

%Receiver ...........................................

%AWGN addition .......................................

[ich4,qch4] = comb(ich3,qch3,attn);

%Guard interval removal ..............................

[ich5,qch5] = girem (ich4,qch4,dctlen2,gilen,nd);
%DCT .................................................
rx = ich5 + qch5.*j;

%%%%%%%%%%% Expanding
rx_lfdma_Abs = abs(rx);
r_lfdma_max = tx_lfdma_max;

RxSamples_lfdma = 
(r_lfdma_max/u)*(exp(log10(1+u)*2.30
2585093*rx_lfdma_Abs./r_lfdma_max)-
1).*sign(rx);

rx_ifdma_Abs = abs(rx);
r_ifdma_max = tx_ifdma_max;

rxx = (r_ifdma_max/u)*(exp(log10(1+u)*2.302585093
*rx_ifdma_Abs./r_ifdma_max)-1).*sign(rx);

ry = dct(rxx);
ich6 = real (ry);
 App en d i x B 3 31

qch6 = imag (ry);

%Demodulation ........................................

ich7 = ich6./kmod;
qch7 = qch6./kmod;
outdemodata = qpskdemod (ich7,qch7,para,nd,m1);

function outdemodata = dwt_channel_clipping_companding

(serdata,para,nd,m1,gilen,dctlen,sr,ebno, br)

%serial to parallel conversion .......................

paradata = reshape(serdata,para,nd*m1);
%QPSK modulation .....................................
[ich,qch] = qpskmod(paradata,para,nd,m1);
% [ich0,qch0] = compoversamp(ich01,qch01,length​
(ich01),Ipoint);
kmod = 1/sqrt(2);
ich1 = ich.*kmod;
qch1 = qch.*kmod;
%IDWT ................................................
x = ich1 + qch1.*j;
y = wavelet('D6',-1,x,'zpd'); % Invert 5 stages
%; % 2D wavelet transform
% R = wavelet('2D CDF 9/7',-2,Y); % Recover X from Y%
Forward transform with 5 stages
CR = 4;
clipping_threshold = (10^(CR/10))*sqrt(mean(ab
s(y).^2));
tx_signal_Ang = angle(y);
for ii = 1: length(y)
if y(ii) = = y(ii);
y(ii) = y(ii);

elseif abs(y(ii)).^2> clipping_threshold

y(ii) = 
clipping_threshold.*exp(sqrt​
(-1)*tx_signal_Ang(ii));

end
end
%companding ................................
u = 4;
332 App en d i x B

tx_lfdma_max = max(abs(y(1: nd*para)));

tx_lfdma_Abs = 
abs(y(1: nd*para)); % tx data
amplitude
TxSamples_lfdma1 = 
tx_lfdma_max*((log10(1+u*​
(tx_lfdma_Abs./tx_lfdma_
max)))/log10(u+1)).*sign​
(y(1: nd*para));
tx_ifdma_max = max(abs(y(1: nd*para)));
tx_ifdma_Abs = 
abs(y(1: nd*para)); % tx data
amplitude
yy = 
tx_ifdma_max*((log10(1+u*(tx_ifdma_Abs./
tx_ifdma_max)))/log10(u+1)).*sign(y(1:
nd*para));

ich2 = real (yy);

qch2 = imag (yy);

%Guard interval insertion .............................

[ich3,qch3] = giins(ich2,qch2,dctlen,gilen,nd);
dctlen2 = dctlen + gilen;

%Attenuation Calculation .............................

spow = sum(ich3.^2+qch3.^2)/nd./para;
attn = 0.5*spow*sr/br*10.^(-ebno/10);
attn = sqrt (attn);

%Receiver ...........................................
%AWGN addition .......................................
[ich4,qch4] = comb(ich3,qch3,attn);

%Guard interval removal ...............................

[ich5,qch5] = girem (ich4,qch4,dctlen2,gilen,nd);

%DWT .................................................
rx = ich5 + qch5.*j;

%%%%%%%%%%% Expanding
rx_lfdma_Abs = abs(rx);
r_lfdma_max = tx_lfdma_max;

RxSamples_lfdma = 
(r_lfdma_max/u)*(exp(log10(1+u)*2.30
2585093*rx_lfdma_Abs./r_lfdma_max)​
-1).*sign(rx);
 App en d i x B 333

rx_ifdma_Abs = abs(rx);
r_ifdma_max = tx_ifdma_max;

rxx = (
r_ifdma_max/u)*(exp(log10(1+u)*2.302585093
*rx_ifdma_Abs./r_ifdma_max)-1).*sign(rx);

ry = wavelet('D6',1,rxx,'zpd');; % 2D wavelet transform

% R = wavelet('2D CDF 9/7',-2,Y); % Recover X from Y;
ich6 = real (ry);
qch6 = imag (ry);

%Demodulation ........................................
ich7 = ich6./kmod;
qch7 = qch6./kmod;
outdemodata = qpskdemod (ich7,qch7,para,nd,m1);

function outdemodata = fft_channel_estimation_no_­

mapping_clipping(serdata,para,nd,m1,gilen,fftlen,sr,e
bno, br,fd,flat)

paradata = reshape(serdata,para,nd*m1);
%QPSK modulation .....................................
[ich,qch] = qpskmod(paradata,para,nd,m1);
% [ich0,qch0] = compoversamp(ich01,qch01,length​
(ich01),Ipoint);
kmod = 1/sqrt(2);
ich1 = ich.*kmod;
qch1 = qch.*kmod;
%channel estimation data generation ..................
kndata = zeros(1,fftlen);
kndata0 = 2.*(rand(1,para)<0.5)-1;
kndata(1: para/2) = kndata0(1: para/2);
kndata((para/2)+1: para) = kndata0((para/2)+1: para);
ceich = kndata;
ceqch = zeros(1,para);
%data mapping ........................................
ich2 = [ceich.' ich1];
qch2 = [ceqch.' qch1];
%IFFT ................................................
x = ich2 + qch2.*i;

y = ifft(x);
CR = 4;
clipping_threshold = 
(10^(CR/10))*sqrt(mean(ab
s(y).^2));
334 App en d i x B

tx_signal_Ang = angle(y);
for ii = 1: length(y)
if y(ii) = = y(ii);
y(ii) = y(ii);

elseif abs(y(ii)).^2> clipping_threshold

y(ii) = 
clipping_threshold.*exp(sqrt(-1)*tx_
signal_Ang(ii));

end
end

ich3 = real (y);

qch3 = imag (y);

%Guard interval insertion .............................

[ich4,qch4] = giins(ich3,qch3,fftlen,gilen,nd+1);
fftlen2 = fftlen + gilen;

%Attenuation Calculation .............................

spow = sum(ich4.^2+qch4.^2)/nd./para;
attn = 0.5*spow*sr/br*10.^(-ebno/10);
attn = sqrt (attn);
%fading ..............................................
%******* Create Rayleigh fading channel object.*******
tstp = 1/sr/(fftlen+gilen);
itau = [0,2,3,4];
dlvll = [0,10,20,25];
n0 = [6,7,6,7];
th1 = [0,0,0,0];
itnd1 = [1000,2000,3000,4000];
now1 = 4;
itnd0 = nd*(fftlen+gilen)*20;
[ifade,qfade,ramp,rcos,rsin] = sefade(ich4,qch4,itau,d
lvll,th1,n0,itnd1,now1,length(ich4),tstp,fd,flat);
itnd1 = itnd1+itnd0;
ich4 = ifade;
qch4 = qfade;

%******************* AWGN addition *******************

%Receiver ...........................................

%AWGN addition .......................................

[ich5,qch5] = comb(ich4,qch4,attn);
 App en d i x B 335

%perfect fading compensation .........................

ifade2 = 1
./ramp.*(rcos(1,: ).*ich5+rsin​
(1,: ).*qch5);
qfade2 = 1
./ramp.*(-rsin(1,: ).*ich5+rcos​
(1,:).*qch5);
ich5 = ifade2;
qch5 = qfade2;

%Guard interval removal ..............................

[ich6,qch6] = girem (ich5,qch5,fftlen2,gilen,nd+1);
%FFT .................................................
rx = ich6 + qch6.*j;
ry = fft(rx);
ich7 = real (ry);
qch7 = imag (ry);
%fading compensation by channel estimation symbol ....
ce = 1;
ice0 = ich2(:,ce);
qce0 = qch2(:,ce);
ice1 = ich7(:,ce);
qce1 = qch7(:,ce);
%calculate reverse rotation ..........................
iv = real((1./(ice1.^2+qce1.^2)).*(ice0+i.*qce0).​
*(ice1-i.*qce1));
qv = imag((1./(ice1.^2+qce1.^2)).*(ice0+i.*qce0).​
*(ice1-i.*qce1));
%matrix for reverse rotation .........................
ieqv1 = [iv iv iv iv iv iv iv];
qeqv1 = [qv qv qv qv qv qv qv];
%reverse rotation ............... ........
icompen = real((ich7+i.*qch7).*(ieqv1+i.*qeqv1));
qcompen = imag((ich7+i.*qch7).*(ieqv1+i.*qeqv1));
ich7 = icompen;
qch7 = qcompen;
%channel estimation symbol removal ...................
knd = 1;  %number of known channel estimation ofdm
symbol
ich9 = ich7(:,knd+1: nd+1);
qch9 = qch7(:,knd+1: nd+1);

%Demodulation ........................................
ich10 = ich9./kmod;
qch10 = qch9./kmod;
outdemodata = qpskdemod (ich10,qch10,para,nd,m1);
336 App en d i x B

function outdemodata = 
dct_channel_estimation_no_
mapping_clipping(serdata,para,n
d,m1,gilen,fftlen,sr,ebno,
br,fd,flat)

paradata = reshape(serdata,para,nd*m1);

%QPSK modulation .....................................

[ich,qch] = qpskmod(paradata,para,nd,m1);
% [ich0,qch0] = 
compoversamp(ich01,qch01,length(ich01)
,Ipoint);
kmod = 1/sqrt(2);
ich1 = ich.*kmod;
qch1 = qch.*kmod;
%channel estimation data generation ..................
kndata = zeros(1,fftlen);
kndata0 = 2.*(rand(1,para)<0.5)-1;
kndata(1: para/2) = kndata0(1: para/2);
kndata((para/2)+1: para) = kndata0((para/2)+1: para);
ceich = kndata;
ceqch = zeros(1,para);
%data mapping ........................................
ich2 = [ceich.' ich1];
qch2 = [ceqch.' qch1];
%IDCT ................................................
x = ich2 + qch2.*i;

y = idct(x);

CR = 4;

clipping_threshold = (10^(CR/10))*sqrt(mean(ab
s(y).^2));
tx_signal_Ang = angle(y);
for ii = 1: length(y)
if y(ii) = = y(ii);
y(ii) = y(ii);
elseif abs(y(ii)).^2> clipping_threshold
y(ii) = clipping_threshold.*exp(sqrt(-1)*tx_
signal_Ang(ii));

end
end

ich3 = real (y);

 App en d i x B 337

qch3 = imag (y);

%Guard interval insertion ............................

[ich4,qch4] = giins(ich3,qch3,fftlen,gilen,nd+1);
fftlen2 = fftlen + gilen;

%Attenuation Calculation .............................

spow = sum(ich4.^2+qch4.^2)/nd./para;
attn = 0.5*spow*sr/br*10.^(-ebno/10);
attn = sqrt (attn);
%fading ..............................................
%******* Create Rayleigh fading channel object.*******
tstp = 1/sr/(fftlen+gilen);
itau = [0,2,3,4];
dlvll = [0,10,20,25];
n0 = [6,7,6,7];
th1 = [0,0,0,0];
itnd1 = [1000,2000,3000,4000];
now1 = 4;
itnd0 = nd*(fftlen+gilen)*20;
[ifade,qfade,ramp,rcos,rsin] = sefade(ich4,qch4,itau,d
lvll,th1,n0,itnd1,now1,length(ich4),tstp,fd,flat);
itnd1 = itnd1+itnd0;
ich4 = ifade;
qch4 = qfade;

%******************* AWGN addition *******************

%Receiver ...........................................

%AWGN addition .......................................

[ich5,qch5] = comb(ich4,qch4,attn);
%perfect fading compensation ...............
............... ..........
ifade2 = 1./ramp.*(rcos(1,: ).*ich5+rsin(1,: ).*qch5);
qfade2 = 1./ramp.*(-rsin(1,: ).*ich5+rcos(1,: ).*qch5);
ich5 = ifade2;
qch5 = qfade2;

%Guard interval removal ..............................

[ich6,qch6] = girem (ich5,qch5,fftlen2,gilen,nd+1);
%DCT .................................................
rx = ich6 + qch6.*j;
338 App en d i x B

ry = dct(rx);
ich7 = real (ry);
qch7 = imag (ry);
%fading compensation by channel estimation symbol ....
ce = 1;
ice0 = ich2(:,ce);
qce0 = qch2(:,ce);
ice1 = ich7(:,ce);
qce1 = qch7(:,ce);
%calculate reverse rotation ..........................
iv = real((1./(ice1.^2+qce1.^2)).*(ice0+i.*qce0).​
*(ice1-i.*qce1));
qv = imag((1./(ice1.^2+qce1.^2)).*(ice0+i.*qce0).​
*(ice1-i.*qce1));
%matrix for reverse rotation .........................
ieqv1 = [iv iv iv iv iv iv iv];
qeqv1 = [qv qv qv qv qv qv qv];
%reverse rotation ............... ........
icompen = real((ich7+i.*qch7).*(ieqv1+i.*qeqv1));
qcompen = imag((ich7+i.*qch7).*(ieqv1+i.*qeqv1));
ich7 = icompen;
qch7 = qcompen;
%channel estimation symbol removal ...................
knd = 1;  %number of known channel
estimation ofdm symbol
ich9 = ich7(:,knd+1: nd+1);
qch9 = qch7(:,knd+1: nd+1);

%Demodulation ........................................
ich10 = ich9./kmod;
qch10 = qch9./kmod;
outdemodata = qpskdemod (ich10,qch10,para,nd,m1);

function outdemodata = dwt_channel_estimation_no_­

mapping_clipping(serdata,para,nd,m1,gilen,fftlen,sr,e
bno, br,fd,flat)

paradata = reshape(serdata,para,nd*m1);
%QPSK modulation .....................................
[ich,qch] = qpskmod(paradata,para,nd,m1);
% [ich0,qch0] = 
compoversamp(ich01,qch01,length​
(ich01),Ipoint);
kmod = 1/sqrt(2);
ich1 = ich.*kmod;
qch1 = qch.*kmod;
 App en d i x B 339

%channel estimation data generation ..................

kndata = zeros(1,fftlen);
kndata0 = 2.*(rand(1,para)<0.5)-1;
kndata(1: para/2) = kndata0(1: para/2);
kndata((para/2)+1: para) = kndata0((para/2)+1: para);
ceich = kndata;
ceqch = zeros(1,para);
%data mapping ........................................
ich2 = [ceich.' ich1];
qch2 = [ceqch.' qch1];

%IDWT ................................................
x = ich2 + qch2.*j;
y = wavelet('D6',-1,x,'zpd'); % Invert 5 stages
%; % 2D wavelet transform
% R = wavelet('2D CDF 9/7',-2,Y); % Recover X from Y%
Forward transform with 5 stages
CR = 4;
clipping_threshold = (10^(CR/10))*sqrt(mean(ab
s(y).^2));
tx_signal_Ang = angle(y);
for ii = 1: length(y)
if y(ii) = = y(ii);
y(ii) = y(ii);
elseif abs(y(ii)).^2> clipping_threshold
y(ii) = 
clipping_threshold.*exp(sqrt​
(-1)*tx_signal_Ang(ii));

end
end

ich3 = real (y);

qch3 = imag (y);
%Guard interval insertion .............................

[ich4,qch4] = giins(ich3,qch3,fftlen,gilen,nd+1);
fftlen2 = fftlen + gilen;

%Attenuation Calculation .............................

spow = sum(ich4.^2+qch4.^2)/nd./para;
attn = 0.5*spow*sr/br*10.^(-ebno/10);
attn = sqrt (attn);
%fading ..............................................
%******* Create Rayleigh fading channel object.*******
340 App en d i x B

tstp = 1/sr/(fftlen+gilen);
itau = [0,2,3,4];
dlvll = [0,10,20,25];
n0 = [6,7,6,7];
th1 = [0,0,0,0];
itnd1 = [1000,2000,3000,4000];
now1 = 4;
itnd0 = nd*(fftlen+gilen)*20;
[ifade,qfade,ramp,rcos,rsin] = sefade(ich4,qch4,itau,d
lvll,th1,n0,itnd1,now1,length(ich4),tstp,fd,flat);
itnd1 = itnd1+itnd0;
ich4 = ifade;
qch4 = qfade;

%*******************AWGN addition ********************

%Receiver ...........................................

%AWGN addition .......................................

[ich5,qch5] = comb(ich4,qch4,attn);
%perfect fading compensation ........................
ifade2 = 1
./ramp.*(rcos(1,: ).*ich5+rsin​
(1,: ).*qch5);
qfade2 = 1
./ramp.*(-rsin(1,: ).*ich5+rcos​
(1,:).*qch5);
ich5 = ifade2;
qch5 = qfade2;

%Guard interval removal ..............................

[ich6,qch6] = girem (ich5,qch5,fftlen2,gilen,nd+1);
%DWT .................................................
rx = ich6 + qch6.*j;
ry = wavelet('D6',1,rx,'zpd');; % 2D wavelet
transform
% R = wavelet('2D CDF 9/7',-2,Y); % Recover X from Y;
ich7 = real (ry);
qch7 = imag (ry);
%fading compensation by channel estimation symbol ....
ce = 1;
ice0 = ich2(:,ce);
qce0 = qch2(:,ce);
ice1 = ich7(:,ce);
qce1 = qch7(:,ce);
%calculate reverse rotation ..........................
iv = real((1./(ice1.^2+qce1.^2)).*(ice0+i.*qce0).​
*(ice1-i.*qce1));
 App en d i x B 3 41

qv = i
mag((1./(ice1.^2+qce1.^2)).*(ice0+i.*qce0).​
*(ice1-i.*qce1));
%matrix for reverse rotation ..........................
ieqv1 = [iv iv iv iv iv iv iv];
qeqv1 = [qv qv qv qv qv qv qv];
%reverse rotation ............... ........
icompen = real((ich7+i.*qch7).*(ieqv1+i.*qeqv1));
qcompen = imag((ich7+i.*qch7).*(ieqv1+i.*qeqv1));
ich7 = icompen;
qch7 = qcompen;
%channel estimation symbol removal ...................
knd = 1; %number of known channel estimation ofdm
symbol
ich9 = ich7(:,knd+1: nd+1);
qch9 = qch7(:,knd+1: nd+1);

%Demodulation ........................................
ich10 = ich9./kmod;
qch10 = qch9./kmod;
outdemodata = qpskdemod (ich10,qch10,para,nd,m1);

function outdemodata = fft_channel_offset_g(serdata,pa

ra,nd,m1,gilen,fftlen,sr,ebno, br,epsilon)

%serial to parallel conversion .......................

paradata = reshape(serdata,para,nd*m1);

%QPSK modulation .....................................

[ich,qch] = qpskmod(paradata,para,nd,m1);

% [ich0,qch0] = c
ompoversamp(ich01,qch01,length​
(ich01),Ipoint);
kmod = 1/sqrt(2);
ich1 = ich.*kmod;
qch1 = qch.*kmod;

%IFFT ................................................
x = ich1 + qch1.*j;
y = ifft(x);
ich2 = real (y);
qch2 = imag (y);

%Guard interval insertion ............................

[ich3,qch3] = giins1(ich2,qch2,fftlen,gilen,nd);
fftlen2 = fftlen + gilen;
3 42 App en d i x B

%Attenuation Calculation ..............................

spow = sum(ich3.^2+qch3.^2)/nd./para;
attn = 0.5*spow*sr/br*10.^(-ebno/10);
attn = sqrt (attn);

%offset ..............................................
n = para/2;
offset = exp(j*2*pi*n*epsilon/para);
i_rx_signal = ich3.*offset;
q_rx_signal = qch3.*offset;
ich3a = i_rx_signal ;
qch3a = q_rx_signal ;

%Receiver ...........................................
%AWGN addition .......................................
[ich4,qch4] = comb(ich3a,qch3a,attn);

%Guard interval removal ..............................

[ich5,qch5] = girem1 (ich4,qch4,fftlen2,gilen,nd);

%FFT .................................................
rx = ich5 + qch5.*j;
ry = fft(rx);
ich6 = real (ry);
qch6 = imag (ry);

%Demodulation ........................................
ich7 = ich6./kmod;
qch7 = qch6./kmod;
outdemodata = qpskdemod (ich7,qch7,para,nd,m1);

function outdemodata = dct_channel_offset_g(serdata,pa

ra,nd,m1,gilen,fftlen,sr,ebno, br,epsilon)

%serial to parallel conversion .......................

paradata = reshape(serdata,para,nd*m1);

%QPSK modulation .....................................

[ich,qch] = qpskmod(paradata,para,nd,m1);

% [ich0,qch0] = c
ompoversamp(ich01,qch01,length​
(ich01),Ipoint);
kmod = 1/sqrt(2);
ich1 = ich.*kmod;
qch1 = qch.*kmod;
 App en d i x B 343

%IDCT ................................................
x = ich1 + qch1.*j;
y = idct(x);
ich2 = real (y);
qch2 = imag (y);

%Guard interval insertion ............................

[ich3,qch3] = giins1(ich2,qch2,fftlen,gilen,nd);
fftlen2 = fftlen + gilen;

%Attenuation Calculation .............................

spow = sum(ich3.^2+qch3.^2)/nd./para;
attn = 0.5*spow*sr/br*10.^(-ebno/10);
attn = sqrt (attn);

%offset ..............................................
n = para/2;
offset = exp(j*2*pi*n*epsilon/para);
i_rx_signal = ich3.*offset;
q_rx_signal = qch3.*offset;
ich3a = i_rx_signal ;
qch3a = q_rx_signal ;

%Receiver ...........................................
%AWGN addition .......................................
[ich4,qch4] = comb(ich3a,qch3a,attn);

%Guard interval removal ...............................

[ich5,qch5] = girem1 (ich4,qch4,fftlen2,gilen,nd);

%DCT .................................................
rx = ich5 + qch5.*j;
ry = dct(rx);
ich6 = real (ry);
qch6 = imag (ry);

%Demodulation ........................................
ich7 = ich6./kmod;
qch7 = qch6./kmod;
outdemodata = qpskdemod (ich7,qch7,para,nd,m1);
function outdemodata = 
dwt_channel_offset_g(serdata,pa
ra,nd,m1,gilen,dctlen,sr,ebno,
br,epsilon)

%serial to parallel conversion .......................

344 App en d i x B

paradata = reshape(serdata,para,nd*m1);
%QPSK modulation .....................................
[ich,qch] = qpskmod(paradata,para,nd,m1);
% [ich0,qch0] = compoversamp(ich01,qch01,length​
(ich01),Ipoint);
kmod = 1/sqrt(2);
ich1 = ich.*kmod;
qch1 = qch.*kmod;
%IDWT ................................................
x = ich1 + qch1.*j;
y = wavelet('D6',-1,x,'zpd'); % Invert 5 stages

% 2D wavelet transf
% R = w
avelet('2D CDF 9/7',-2,Y); % Recover X from Y%
Forward transform with 5 stages
ich2 = real (y);
qch2 = imag (y);

%Guard interval insertion ............................

[ich3,qch3] = giins1(ich2,qch2,dctlen,gilen,nd);
dctlen2 = dctlen + gilen;

%Attenuation Calculation .............................

spow = sum(ich3.^2+qch3.^2)/nd./para;
attn = 0.5*spow*sr/br*10.^(-ebno/10);
attn = sqrt (attn);

%offset ..............................................
n = para/2;
offset = exp(j*2*pi*n*epsilon/para);
i_rx_signal = ich3.*offset;
q_rx_signal = qch3.*offset;
ich3a = i_rx_signal ;
qch3a = q_rx_signal ;

%Receiver ...........................................
%AWGN addition .......................................
[ich4,qch4] = comb(ich3a,qch3a,attn);
%Guard interval removal ..............................
[ich5,qch5] = girem1 (ich4,qch4,dctlen2,gilen,nd);
%DWT .................................................
rx = ich5 + qch5.*j;
ry = wavelet('D6',1,rx,'zpd');; % 2D wavelet transform
% R = wavelet('2D CDF 9/7',-2,Y); % Recover X from Y;
ich6 = real (ry);
 App en d i x B 345

qch6 = imag (ry);

%Demodulation ........................................
ich7 = ich6./kmod;
qch7 = qch6./kmod;
outdemodata = qpskdemod (ich7,qch7,para,nd,m1);

function outdemodata = 
fft_channel_offset_zg(serdata,p
ara,nd,m1,gilen,fftlen,sr,ebno,
br,epsilon)

%serial to parallel conversion .......................

paradata = reshape(serdata,para,nd*m1);

%QPSK modulation .....................................

[ich,qch] = qpskmod(paradata,para,nd,m1);

% [ich0,qch0] = c
ompoversamp(ich01,qch01,length​
(ich01),Ipoint);
kmod = 1/sqrt(2);
ich1 = ich.*kmod;
qch1 = qch.*kmod;

%IFFT ................................................
x = ich1 + qch1.*j;
y = ifft(x);
ich2 = real (y);
qch2 = imag (y);

%Guard interval insertion ............................

[ich3,qch3] = giins2(ich2,qch2,fftlen,gilen,nd);
fftlen2 = fftlen + gilen;

%Attenuation Calculation .............................

spow = sum(ich3.^2+qch3.^2)/nd./para;
attn = 0.5*spow*sr/br*10.^(-ebno/10);
attn = sqrt (attn);

%offset ..............................................
n = para/2;
offset = exp(j*2*pi*n*epsilon/para);
i_rx_signal = ich3.*offset;
q_rx_signal = qch3.*offset;
ich3a = i_rx_signal ;
qch3a = q_rx_signal ;
346 App en d i x B

%Receiver ...........................................
%AWGN addition .......................................
[ich4,qch4] = comb(ich3a,qch3a,attn);

%Guard interval removal ..............................

[ich5,qch5] = girem1 (ich4,qch4,fftlen2,gilen,nd);

%FFT .................................................
rx = ich5 + qch5.*j;
ry = fft(rx);
ich6 = real (ry);
qch6 = imag (ry);

%Demodulation ........................................
ich7 = ich6./kmod;
qch7 = qch6./kmod;
outdemodata = qpskdemod (ich7,qch7,para,nd,m1);

function outdemodata = 
dct_channel_offset_zg(serdata,p
ara,nd,m1,gilen,fftlen,sr,ebno,
br,epsilon)

%serial to parallel conversion .......................

paradata = reshape(serdata,para,nd*m1);

%QPSK modulation .....................................

[ich,qch] = qpskmod(paradata,para,nd,m1);

% [ich0,qch0] = c
ompoversamp(ich01,qch01,length​
(ich01),Ipoint);
kmod = 1/sqrt(2);
ich1 = ich.*kmod;
qch1 = qch.*kmod;

%IDCT ................................................
x = ich1 + qch1.*j;
y = idct(x);
ich2 = real (y);
qch2 = imag (y);

%Guard interval insertion ............................

[ich3,qch3] = giins2(ich2,qch2,fftlen,gilen,nd);
fftlen2 = fftlen + gilen;

%Attenuation Calculation .............................

 App en d i x B 3 47

spow = sum(ich3.^2+qch3.^2)/nd./para;
attn = 0.5*spow*sr/br*10.^(-ebno/10);
attn = sqrt (attn);

%offset ..............................................
n = para/2;
offset = exp(j*2*pi*n*epsilon/para);
i_rx_signal = ich3.*offset;
q_rx_signal = qch3.*offset;
ich3a = i_rx_signal ;
qch3a = q_rx_signal ;

%Receiver ...........................................
%AWGN addition .......................................
[ich4,qch4] = comb(ich3a,qch3a,attn);

%Guard interval removal ...............................

[ich5,qch5] = girem1 (ich4,qch4,fftlen2,gilen,nd);

%DCT .................................................
rx = ich5 + qch5.*j;
ry = dct(rx);
ich6 = real (ry);
qch6 = imag (ry);

%Demodulation ........................................
ich7 = ich6./kmod;
qch7 = qch6./kmod;
outdemodata = qpskdemod (ich7,qch7,para,nd,m1);

function outdemodata = 
dwt_channel_offset_zg(serdata,p
ara,nd,m1,gilen,dctlen,sr,ebno,
br,epsilon)

%serial to parallel conversion .......................

paradata = reshape(serdata,para,nd*m1);
%QPSK modulation .....................................
[ich,qch] = qpskmod(paradata,para,nd,m1);
% [ich0,qch0] = 
compoversamp(ich01,qch01,length​
(ich01),Ipoint);
kmod = 1/sqrt(2);
ich1 = ich.*kmod;
qch1 = qch.*kmod;
%IDWT ................................................
x = ich1 + qch1.*j;
348 App en d i x B

y = wavelet('D6',-1,x,'zpd'); % Invert 5 stages

% 2D wavelet transf
% R = wavelet('2D CDF 9/7',-2,Y); % Recover X from Y%
Forward transform with 5 stages
ich2 = real (y);
qch2 = imag (y);

%Guard interval insertion ............................

[ich3,qch3] = giins2(ich2,qch2,dctlen,gilen,nd);
dctlen2 = dctlen + gilen;

%Attenuation Calculation .............................

spow = sum(ich3.^2+qch3.^2)/nd./para;
attn = 0.5*spow*sr/br*10.^(-ebno/10);
attn = sqrt (attn);

%offset ..............................................
n = para/2;
offset = exp(j*2*pi*n*epsilon/para);
i_rx_signal = ich3.*offset;
q_rx_signal = qch3.*offset;
ich3a = i_rx_signal ;
qch3a = q_rx_signal ;

%Receiver ...........................................
%AWGN addition .......................................
[ich4,qch4] = comb(ich3a,qch3a,attn);
%Guard interval removal ..............................
[ich5,qch5] = girem1 (ich4,qch4,dctlen2,gilen,nd);
%DWT .................................................
rx = ich5 + qch5.*j;
ry = wavelet('D6',1,rx,'zpd');; % 2D wavelet transform
% R = wavelet('2D CDF 9/7',-2,Y); % Recover X from Y;
ich6 = real (ry);
qch6 = imag (ry);

%Demodulation ........................................
ich7 = ich6./kmod;
qch7 = qch6./kmod;
outdemodata = qpskdemod (ich7,qch7,para,nd,m1);
function outdemodata = 
fft_channel_estimation_no_mappi
ng_g(serdata,para,nd,m1,gilen,f
ftlen,sr,ebno, br,fd,flat)
 App en d i x B 349

paradata = reshape(serdata,para,nd*m1);
%QPSK modulation .....................................
[ich,qch] = qpskmod(paradata,para,nd,m1);
% [ich0,qch0] = compoversamp(ich01,qch01,length(ich01)
,Ipoint);
kmod = 1/sqrt(2);
ich1 = ich.*kmod;
qch1 = qch.*kmod;
%channel estimation data generation ..................
kndata = zeros(1,fftlen);
kndata0 = 2.*(rand(1,para)<0.5)-1;
kndata(1: para/2) = kndata0(1: para/2);
kndata((para/2)+1: para) = kndata0((para/2)+1: para);
ceich = kndata;
ceqch = zeros(1,para);
%data mapping ........................................
ich2 = [ceich.' ich1];
qch2 = [ceqch.' qch1];

%IFFT ................................................
x = ich2 + qch2.*i;
y = ifft(x);
ich3 = real (y);
qch3 = imag (y);

%Guard interval insertion .............................

[ich4,qch4] = giins1(ich3,qch3,fftlen,gilen,nd+1);
fftlen2 = fftlen + gilen;

%Attenuation Calculation .............................

spow = sum(ich4.^2+qch4.^2)/nd./para;
attn = 0.5*spow*sr/br*10.^(-ebno/10);
attn = sqrt (attn);

%fading ..............................................
%******* Create Rayleigh fading channel object.*******
tstp = 1/sr/(fftlen+gilen);
itau = [0,2,3,4];
dlvll = [0,10,20,25];
n0 = [6,7,6,7];
th1 = [0,0,0,0];
itnd1 = [1000,2000,3000,4000];
now1 = 4;
itnd0 = nd*(fftlen+gilen)*20;
350 App en d i x B

[ifade,qfade,ramp,rcos,rsin] = sefade(ich4,qch4,itau,d
lvll,th1,n0,itnd1,now1,length(ich4),tstp,fd,flat);
itnd1 = itnd1+itnd0;
ich4 = ifade;
qch4 = qfade;

%******************* AWGN addition *******************

%Receiver ...........................................
%AWGN addition .......................................
[ich5,qch5] = comb(ich4,qch4,attn);
%perfect fading compensation ..............
ifade2 = 1
./ramp.*(rcos(1,: ).*ich5+rsin​
(1,: ).*qch5);
qfade2 = 1
./ramp.*(-rsin(1,: ).*ich5+rcos​
(1,: ).*qch5);
ich5 = ifade2;
qch5 = qfade2;

%Guard interval removal ..............................

[ich6,qch6] = girem1 (ich5,qch5,fftlen2,gilen,nd+1);
%FFT .................................................
rx = ich6 + qch6.*j;
ry = fft(rx);
ich7 = real (ry);
qch7 = imag (ry);
%fading compensation by channel estimation symbol ....
ce = 1;
ice0 = ich2(:,ce);
qce0 = qch2(:,ce);
ice1 = ich7(:,ce);
qce1 = qch7(:,ce);
%calculate reverse rotation ..........................
iv = real((1./(ice1.^2+qce1.^2)).*(ice0+i.*qce0).​
*(ice1-i.*qce1));
qv = imag((1./(ice1.^2+qce1.^2)).*(ice0+i.*qce0).​
*(ice1-i.*qce1));
%matrix for reverse rotation .........................
ieqv1 = [iv iv iv iv iv iv iv];
qeqv1 = [qv qv qv qv qv qv qv];
%reverse rotation ........................
icompen = real((ich7+i.*qch7).*(ieqv1+i.*qeqv1));
qcompen = imag((ich7+i.*qch7).*(ieqv1+i.*qeqv1));
ich7 = icompen;
qch7 = qcompen;
%channel estimation symbol removal ...................
 App en d i x B 3 51

knd = 1; %number of known channel estimation ofdm

symbol
ich9 = ich7(:,knd+1: nd+1);
qch9 = qch7(:,knd+1: nd+1);

%Demodulation ........................................
ich10 = ich9./kmod;
qch10 = qch9./kmod;
outdemodata = qpskdemod (ich10,qch10,para,nd,m1);

function outdemodata = 
dct_channel_estimation_no_mappi
ng_g(serdata,para,nd,m1,gilen,f
ftlen,sr,ebno, br,fd,flat)

paradata = reshape(serdata,para,nd*m1);
%QPSK modulation .....................................
[ich,qch] = qpskmod(paradata,para,nd,m1);
% [ich0,qch0] = 
compoversamp(ich01,qch01,length​
(ich01),Ipoint);
kmod = 1/sqrt(2);
ich1 = ich.*kmod;
qch1 = qch.*kmod;
%channel estimation data generation ..................
kndata = zeros(1,fftlen);
kndata0 = 2.*(rand(1,para)<0.5)-1;
kndata(1: para/2) = kndata0(1: para/2);
kndata((para/2)+1: para) = kndata0((para/2)+1: para);
ceich = kndata;
ceqch = zeros(1,para);
%data mapping ........................................
ich2 = [ceich.' ich1];
qch2 = [ceqch.' qch1];

%IDCT ................................................
x = ich2 + qch2.*i;
y = idct(x);
ich3 = real (y);
qch3 = imag (y);

%Guard interval insertion .............................

[ich4,qch4] = giins1(ich3,qch3,fftlen,gilen,nd+1);
fftlen2 = fftlen + gilen;

%Attenuation Calculation .............................

352 App en d i x B

spow = sum(ich4.^2+qch4.^2)/nd./para;
attn = 0.5*spow*sr/br*10.^(-ebno/10);
attn = sqrt (attn);
%fading ..............................................
%******* Create Rayleigh fading channel object.*******
tstp = 1/sr/(fftlen+gilen);
itau = [0,2,3,4];
dlvll = [0,10,20,25];
n0 = [6,7,6,7];
th1 = [0,0,0,0];
itnd1 = [1000,2000,3000,4000];
now1 = 4;
itnd0 = nd*(fftlen+gilen)*20;
[ifade,qfade,ramp,rcos,rsin] = sefade(ich4,qch4,itau,d
lvll,th1,n0,itnd1,now1,length(ich4),tstp,fd,flat);
itnd1 = itnd1+itnd0;
ich4 = ifade;
qch4 = qfade;

%****************** AWGN addition ********************

%Receiver ...........................................

%AWGN addition .......................................

[ich5,qch5] = comb(ich4,qch4,attn);
%perfect fading compensation ........................
ifade2 = 1
./ramp.*(rcos(1,: ).*ich5+rsin​
(1,: ).*qch5);
qfade2 = 1
./ramp.*(-rsin(1,: ).*ich5+rcos​
(1,: ).*qch5);
ich5 = ifade2;
qch5 = qfade2;

%Guard interval removal ..............................

[ich6,qch6] = girem1 (ich5,qch5,fftlen2,gilen,nd+1);
%DCT .................................................
rx = ich6 + qch6.*j;
ry = dct(rx);
ich7 = real (ry);
qch7 = imag (ry);
%fading compensation by channel estimation symbol ....
ce = 1;
ice0 = ich2(:,ce);
qce0 = qch2(:,ce);
ice1 = ich7(:,ce);
qce1 = qch7(:,ce);
 App en d i x B 353

%calculate reverse rotation ...........................

iv = real((1./(ice1.^2+qce1.^2)).*(ice0+i.*qce0).​
*(ice1-i.*qce1));
qv = imag((1./(ice1.^2+qce1.^2)).*(ice0+i.*qce0).​
*(ice1-i.*qce1));
%matrix for reverse rotation ..........................
ieqv1 = [iv iv iv iv iv iv iv];
qeqv1 = [qv qv qv qv qv qv qv];
%reverse rotation ........................
icompen = real((ich7+i.*qch7).*(ieqv1+i.*qeqv1));
qcompen = imag((ich7+i.*qch7).*(ieqv1+i.*qeqv1));
ich7 = icompen;
qch7 = qcompen;
%channel estimation symbol removal ...................
knd = 1; %number of known channel estimation ofdm
symbol
ich9 = ich7(:,knd+1: nd+1);
qch9 = qch7(:,knd+1: nd+1);

%Demodulation ........................................
ich10 = ich9./kmod;
qch10 = qch9./kmod;
outdemodata = qpskdemod (ich10,qch10,para,nd,m1);

function outdemodata = 
dwt_channel_estimation_no_mappi
ng_g(serdata,para,nd,m1,gilen,f
ftlen,sr,ebno, br,fd,flat)

paradata = reshape(serdata,para,nd*m1);
%QPSK modulation .....................................
[ich,qch] = qpskmod(paradata,para,nd,m1);
% [ich0,qch0] = compoversamp(ich01,qch01,length(ich01)
,Ipoint);
kmod = 1/sqrt(2);
ich1 = ich.*kmod;
qch1 = qch.*kmod;
%channel estimation data generation ..................
kndata = zeros(1,fftlen);
kndata0 = 2.*(rand(1,para)<0.5)-1;
kndata(1: para/2) = kndata0(1: para/2);
kndata((para/2)+1: para) = kndata0((para/2)+1: para);
ceich = kndata;
ceqch = zeros(1,para);
%data mapping ........................................
ich2 = [ceich.' ich1];
354 App en d i x B

qch2 = [ceqch.' qch1];

%IDWT ................................................
x = ich2 + qch2.*j;
y = wavelet('D6',-1,x,'zpd'); % Invert 5 stages
%; % 2D wavelet transform
% R = wavelet('2D CDF 9/7',-2,Y); % Recover X from Y%
Forward transform with 5 stages

ich3 = real (y);

qch3 = imag (y);

%Guard interval insertion ............................

[ich4,qch4] = giins1(ich3,qch3,fftlen,gilen,nd+1);
fftlen2 = fftlen + gilen;

%Attenuation Calculation .............................

spow = sum(ich4.^2+qch4.^2)/nd./para;
attn = 0.5*spow*sr/br*10.^(-ebno/10);
attn = sqrt (attn);
%fading ..............................................
%******* Create Rayleigh fading channel object.*******
tstp = 1/sr/(fftlen+gilen);
itau = [0,2,3,4];
dlvll = [0,10,20,25];
n0 = [6,7,6,7];
th1 = [0,0,0,0];
itnd1 = [1000,2000,3000,4000];
now1 = 4;
itnd0 = nd*(fftlen+gilen)*20;
[ifade,qfade,ramp,rcos,rsin] = sefade(ich4,qch4,itau,d
lvll,th1,n0,itnd1,now1,length(ich4),tstp,fd,flat);
itnd1 = itnd1+itnd0;
ich4 = ifade;
qch4 = qfade;

%****************** AWGN addition ********************

%Receiver ...........................................
%AWGN addition .......................................
[ich5,qch5] = comb(ich4,qch4,attn);
%perfect fading compensation ........................
ifade2 = 1
./ramp.*(rcos(1,: ).*ich5+rsin​
(1,: ).*qch5);
 App en d i x B 355

qfade2 = 1
./ramp.*(-rsin(1,: ).*ich5+rcos​
(1,: ).*qch5);
ich5 = ifade2;
qch5 = qfade2;

%Guard interval removal ..............................

[ich6,qch6] = 
girem1 (ich5,qch5,fftlen2,gilen,nd+1);
%DWT .................................................
rx = ich6 + qch6.*j;
ry = wavelet('D6',1,rx,'zpd');; % 2D wavelet
transform
% R = wavelet('2D CDF 9/7',-2,Y); % Recover X from Y;
ich7 = real (ry);
qch7 = imag (ry);

%fading compensation by channel estimation symbol ....

ce = 1;
ice0 = ich2(:,ce);
qce0 = qch2(:,ce);
ice1 = ich7(:,ce);
qce1 = qch7(:,ce);
%calculate reverse rotation ..........................
iv = real((1./(ice1.^2+qce1.^2)).*(ice0+i.*qce0).​
*(ice1-i.*qce1));
qv = imag((1./(ice1.^2+qce1.^2)).*(ice0+i.*qce0).​
*(ice1-i.*qce1));
%matrix for reverse rotation ..........................
ieqv1 = [iv iv iv iv iv iv iv];
qeqv1 = [qv qv qv qv qv qv qv];
%reverse rotation ........................
icompen = real((ich7+i.*qch7).*(ieqv1+i.*qeqv1));
qcompen = imag((ich7+i.*qch7).*(ieqv1+i.*qeqv1));
ich7 = icompen;
qch7 = qcompen;
%channel estimation symbol removal ...................
knd = 1; %number of known channel estimation ofdm
symbol
ich9 = ich7(:,knd+1: nd+1);
qch9 = qch7(:,knd+1: nd+1);

%Demodulation ........................................
ich10 = ich9./kmod;
qch10 = qch9./kmod;
outdemodata = qpskdemod (ich10,qch10,para,nd,m1);
356 App en d i x B

function outdemodata = 
fft_channel_estimation_no_map-
ping_zg(serdata,para,nd,m1,gile
n,fftlen,sr,ebno, br,fd,flat)

paradata = reshape(serdata,para,nd*m1);
%QPSK modulation .....................................
[ich,qch] = qpskmod(paradata,para,nd,m1);
% [ich0,qch0] = compoversamp(ich01,qch01,length​
(ich01),Ipoint);
kmod = 1/sqrt(2);
ich1 = ich.*kmod;
qch1 = qch.*kmod;
%channel estimation data generation ..................
kndata = zeros(1,fftlen);
kndata0 = 2.*(rand(1,para)<0.5)-1;
kndata(1: para/2) = kndata0(1: para/2);
kndata((para/2)+1: para) = kndata0((para/2)+1: para);
ceich = kndata;
ceqch = zeros(1,para);
%data mapping ........................................
ich2 = [ceich.' ich1];
qch2 = [ceqch.' qch1];

%IFFT ................................................
x = ich2 + qch2.*i;
y = ifft(x);
ich3 = real (y);
qch3 = imag (y);

%Guard interval insertion ............................

[ich4,qch4] = giins2(ich3,qch3,fftlen,gilen,nd+1);
fftlen2 = fftlen + gilen;

%Attenuation Calculation .............................

spow = sum(ich4.^2+qch4.^2)/nd./para;
attn = 0.5*spow*sr/br*10.^(-ebno/10);
attn = sqrt (attn);
%fading ..............................................
%*******Create Rayleigh fading channel object.********
tstp = 1/sr/(fftlen+gilen);
itau = [0,2,3,4];
dlvll = [0,10,20,25];
n0 = [6,7,6,7];
th1 = [0,0,0,0];
 App en d i x B 357

itnd1 = [1000,2000,3000,4000];
now1 = 4;
itnd0 = nd*(fftlen+gilen)*20;
[ifade,qfade,ramp,rcos,rsin] = sefade(ich4,qch4,itau,d
lvll,th1,n0,itnd1,now1,length(ich4),tstp,fd,flat);
itnd1 = itnd1+itnd0;
ich4 = ifade;
qch4 = qfade;

%***************** AWGN addition *********************

%Receiver ...........................................
%AWGN addition .......................................

[ich5,qch5] = comb(ich4,qch4,attn);
%perfect fading compensation ........................
ifade2 = 1
./ramp.*(rcos(1,: ).​
*ich5+rsin(1,: ).*qch5);
qfade2 = 1./ramp.*(-rsin(1,: ).*ich5+rcos​
(1,:).*qch5);
ich5 = ifade2;
qch5 = qfade2;

%Guard interval removal ...............................

[ich6,qch6] = girem1 (ich5,qch5,fftlen2,gilen,nd+1);
%FFT .................................................
rx = ich6 + qch6.*j;
ry = fft(rx);
ich7 = real (ry);
qch7 = imag (ry);
%fading compensation by channel estimation symbol ....
ce = 1;
ice0 = ich2(:,ce);
qce0 = qch2(:,ce);
ice1 = ich7(:,ce);
qce1 = qch7(:,ce);
%calculate reverse rotation ...........................
iv = real((1./(ice1.^2+qce1.^2)).*(ice0+i.*qce0).
*(ice1-i.*qce1));
qv = imag((1./(ice1.^2+qce1.^2)).*(ice0+i.*qce0).
*(ice1-i.*qce1));
%matrix for reverse rotation ..........................
ieqv1 = [iv iv iv iv iv iv iv];
qeqv1 = [qv qv qv qv qv qv qv];
%reverse rotation ............... ........
icompen = real((ich7+i.*qch7).*(ieqv1+i.*qeqv1));
358 App en d i x B

qcompen = imag((ich7+i.*qch7).*(ieqv1+i.*qeqv1));
ich7 = icompen;
qch7 = qcompen;
%channel estimation symbol removal ...................
knd = 1;  %number of known channel estimation ofdm
symbol
ich9 = ich7(:,knd+1: nd+1);
qch9 = qch7(:,knd+1: nd+1);

%Demodulation ........................................
ich10 = ich9./kmod;
qch10 = qch9./kmod;
outdemodata = qpskdemod (ich10,qch10,para,nd,m1);

function outdemodata = 
dct_channel_estimation_no_map-
ping_zg(serdata,para,nd,m1,gile
n,fftlen,sr,ebno, br,fd,flat)

paradata = reshape(serdata,para,nd*m1);
%QPSK modulation .....................................
[ich,qch] = qpskmod(paradata,para,nd,m1);
% [ich0,qch0] = 
compoversamp(ich01,qch01,length
(ich01),Ipoint);
kmod = 1/sqrt(2);
ich1 = ich.*kmod;
qch1 = qch.*kmod;
%channel estimation data generation ..................
kndata = zeros(1,fftlen);
kndata0 = 2.*(rand(1,para)<0.5)-1;
kndata(1: para/2) = kndata0(1: para/2);
kndata((para/2)+1: para) = kndata0((para/2)+1: para);
ceich = kndata;
ceqch = zeros(1,para);
%data mapping ........................................
ich2 = [ceich.' ich1];
qch2 = [ceqch.' qch1];

%IDCT ................................................
x = ich2 + qch2.*i;
y = idct(x);
ich3 = real (y);
qch3 = imag (y);
%Guard interval insertion ............................
[ich4,qch4] = giins2(ich3,qch3,fftlen,gilen,nd+1);
fftlen2 = fftlen + gilen;
 App en d i x B 359

%Attenuation Calculation .............................

spow = sum(ich4.^2+qch4.^2)/nd./para;
attn = 0.5*spow*sr/br*10.^(-ebno/10);
attn = sqrt (attn);
%fading ..............................................
%******* Create Rayleigh fading channel object.*******
tstp = 1/sr/(fftlen+gilen);
itau = [0,2,3,4];
dlvll = [0,10,20,25];
n0 = [6,7,6,7];
th1 = [0,0,0,0];
itnd1 = [1000,2000,3000,4000];
now1 = 4;
itnd0 = nd*(fftlen+gilen)*20;
[ifade,qfade,ramp,rcos,rsin] = sefade(ich4,qch4,itau,d
lvll,th1,n0,itnd1,now1,length(ich4),tstp,fd,flat);
itnd1 = itnd1+itnd0;
ich4 = ifade;
qch4 = qfade;

%******************* AWGN addition *******************

%Receiver ...........................................

%AWGN addition .......................................

[ich5,qch5] = comb(ich4,qch4,attn);
%perfect fading compensation ........................
ifade2 = 1./ramp.*(rcos(1,: ).
*ich5+rsin​
(1,: ).*qch5);
qfade2 = 1./ramp.*(-rsin(1,: ).
*ich5+rcos​
(1,: ).*qch5);
ich5 = ifade2;
qch5 = qfade2;

%Guard interval removal ..............................

[ich6,qch6] = girem1 (ich5,qch5,fftlen2,gilen,nd+1);
%DCT .................................................
rx = ich6 + qch6.*j;
ry = dct(rx);
ich7 = real (ry);
qch7 = imag (ry);
%fading compensation by channel estimation symbol ....
ce = 1;
ice0 = ich2(:,ce);
qce0 = qch2(:,ce);
360 App en d i x B

ice1 = ich7(:,ce);
qce1 = qch7(:,ce);
%calculate reverse rotation ..........................
iv = real((1./(ice1.^2+qce1.^2)).*(ice0+i.*qce0).​
*(ice1-i.*qce1));
qv = imag((1./(ice1.^2+qce1.^2)).*(ice0+i.*qce0).​
*(ice1-i.*qce1));
%matrix for reverse rotation ..........................
ieqv1 = [iv iv iv iv iv iv iv];
qeqv1 = [qv qv qv qv qv qv qv];
%reverse rotation ............... ........
icompen = real((ich7+i.*qch7).*(ieqv1+i.*qeqv1));
qcompen = imag((ich7+i.*qch7).*(ieqv1+i.*qeqv1));
ich7 = icompen;
qch7 = qcompen;
%channel estimation symbol removal ...................
knd = 1; %number of known channel estimation ofdm
symbol
ich9 = ich7(:,knd+1: nd+1);
qch9 = qch7(:,knd+1: nd+1);

%Demodulation ........................................
ich10 = ich9./kmod;
qch10 = qch9./kmod;
outdemodata = qpskdemod (ich10,qch10,para,nd,m1);

function outdemodata = 
dwt_channel_estimation_no_map-
ping_zg(serdata,para,nd,m1,gile
n,fftlen,sr,ebno, br,fd,flat)

paradata = reshape(serdata,para,nd*m1);
%QPSK modulation .....................................
[ich,qch] = qpskmod(paradata,para,nd,m1);
% [ich0,qch0] = compoversamp(ich01,qch01,length(ich01)
,Ipoint);
kmod = 1/sqrt(2);
ich1 = ich.*kmod;
qch1 = qch.*kmod;
%channel estimation data generation ..................
kndata = zeros(1,fftlen);
kndata0 = 2.*(rand(1,para)<0.5)-1;
kndata(1: para/2) = kndata0(1: para/2);
kndata((para/2)+1: para) = kndata0((para/2)+1: para);
ceich = kndata;
ceqch = zeros(1,para);
 App en d i x B 3 61

%data mapping ........................................

ich2 = [ceich.' ich1];
qch2 = [ceqch.' qch1];

%IDWT ................................................
x = ich2 + qch2.*j;
y = wavelet('D6',-1,x,'zpd'); % Invert 5 stages
%; % 2D wavelet transform
% R = wavelet('2D CDF 9/7',-2,Y); % Recover X from Y%
Forward transform with 5 stages

ich3 = real (y);

qch3 = imag (y);

%Guard interval insertion ............................

[ich4,qch4] = giins2(ich3,qch3,fftlen,gilen,nd+1);
fftlen2 = fftlen + gilen;

%Attenuation Calculation .............................

spow = sum(ich4.^2+qch4.^2)/nd./para;
attn = 0.5*spow*sr/br*10.^(-ebno/10);
attn = sqrt (attn);
%fading ..............................................
%****** Create Rayleigh fading channel object.********
tstp = 1/sr/(fftlen+gilen);
itau = [0,2,3,4];
dlvll = [0,10,20,25];
n0 = [6,7,6,7];
th1 = [0,0,0,0];
itnd1 = [1000,2000,3000,4000];
now1 = 4;
itnd0 = nd*(fftlen+gilen)*20;
[ifade,qfade,ramp,rcos,rsin] = sefade(ich4,qch4,itau,d
lvll,th1,n0,itnd1,now1,length(ich4),tstp,fd,flat);
itnd1 = itnd1+itnd0;
ich4 = ifade;
qch4 = qfade;

%******************* AWGN addition *******************

%Receiver ...........................................

%AWGN addition .......................................

[ich5,qch5] = comb(ich4,qch4,attn);
%perfect fading compensation ........................
362 App en d i x B

ifade2 = 1
./ramp.*(rcos(1,: ).*ich5+rsin​
(1,: ).*qch5);
qfade2 = 
1./ramp.*(-rsin(1,: ).*ich5+rcos​
(1,: ).*qch5);
ich5 = ifade2;
qch5 = qfade2;

%Guard interval removal ...............................

[ich6,qch6] = girem1 (ich5,qch5,fftlen2,gilen,nd+1);
%DWT .................................................
rx = ich6 + qch6.*j;
ry = wavelet('D6',1,rx,'zpd');; % 2D wavelet transform
% R = wavelet('2D CDF 9/7',-2,Y); % Recover X from Y;
ich7 = real (ry);
qch7 = imag (ry);

%fading compensation by channel estimation symbol ....

ce = 1;
ice0 = ich2(:,ce);
qce0 = qch2(:,ce);
ice1 = ich7(:,ce);
qce1 = qch7(:,ce);
%calculate reverse rotation ..........................
iv = real((1./(ice1.^2+qce1.^2)).*(ice0+i.*qce0).​
*(ice1-i.*qce1));
qv = imag((1./(ice1.^2+qce1.^2)).*(ice0+i.*qce0).​
*(ice1-i.*qce1));
%matrix for reverse rotation ..........................
ieqv1 = [iv iv iv iv iv iv iv];
qeqv1 = [qv qv qv qv qv qv qv];
%reverse rotation ............... ........
icompen = real((ich7+i.*qch7).*(ieqv1+i.*qeqv1));
qcompen = imag((ich7+i.*qch7).*(ieqv1+i.*qeqv1));
ich7 = icompen;
qch7 = qcompen;
%channel estimation symbol removal ...................
knd = 1; %number of known channel estimation ofdm
symbol
ich9 = ich7(:,knd+1: nd+1);
qch9 = qch7(:,knd+1: nd+1);

%Demodulation ........................................
ich10 = ich9./kmod;
qch10 = qch9./kmod;
outdemodata = qpskdemod (ich10,qch10,para,nd,m1);
 App en d i x B 363

function [iout,qout] = qpskmod(paradata,para,nd,m1)

%**************** variables **********************

% paradata : input data (para-by-nd matrix)
% iout : output Ich data
% qout : output Qch data
% para : Number of parallel channels
% nd : Number of data
% m1 : Number of modulation levels
% (QPSK = 2 16QAM = 4)

%*****************************************************

m2 = m1./2;

paradata2 = paradata.*2-1;
count2 = 0;
for jjj = 1: nd
isi = zeros (para,1);
isq = zeros (para,1);
for ii = 1 : m2
isi = 
isi + 2.^(m2 - ii).*paradata2
((1: para),ii+count2) ;
isq = 
isq + 2.^(m2 - ii).*paradata2
((1: para),m2+ii+count2) ;
end
iout((1: para),jjj) = isi;
qout((1: para),jjj) = isq;
count2 = count2 + m1;
end
%******************** End of file ********************

function [demodata] = 
qpskdemod(idata,
qdata,para,nd,m1)

%**************** variables **********************

% idata : input Ich data

% qdata : input Qch data
% demodata : demodulated data
% para : Number of parallel channels
% nd : Number of data
% m1 : Number of modulation levels
% (QPSK = 2 16QAM = 4)
%*****************************************************
364 App en d i x B

demodata = zeros (para,m1*nd);

demodata ((1: para), (1: m1 : m1*nd-1)) =
idata((1: para), (1: nd)) > = 0;
demodata ((1: para), (2: m1 : m1*nd)) =
qdata((1: para), (1: nd)) > = 0;

%****************** End of file *********************

function varargout = 
wavelet(WaveletName,Level,X,Ext,
Dim)
%WAVELET Discrete wavelet transform.
% 
Y = WAVELET(W,L,X) computes the L-stage discrete
wavelet transform
% 
(DWT) of signal X using wavelet W. The length of X
must be
% 
divisible by 2^L. For the inverse transform,
WAVELET(W,-L,X)
% inverts L stages. Choices for W are
% 'Haar' Haar
% 'D1','D2','D3','D4','D5','D6' Daubechies'
% 'Sym1','Sym2','Sym3','Sym4','Sym5','Sym6' Symlets
% 'Coif1','Coif2' Coiflets
% 'BCoif1' Coiflet-like [2]
% 'Spline Nr.Nd' (or 'bior Nr.Nd') for Splines
% Nr = 0, Nd = 0,1,2,3,4,5,6,7, or 8
% Nr = 1, Nd = 0,1,3,5, or 7
% Nr = 2, Nd = 0,1,2,4,6, or 8
% Nr = 3, Nd = 0,1,3,5, or 7
% Nr = 4, Nd = 0,1,2,4,6, or 8
% Nr = 5, Nd = 0,1,3, or 5
% 
'RSpline Nr.Nd' for the same Nr.Nd pairs Reverse
splines
% 
'S+P (2,2)','S+P (4,2)','S+P (6,2)', S+P
wavelets [3]
% 'S+P (4,4)','S+P (2+2,2)'
% 'TT' “Two-Ten” [5]
% 
'LC 5/3','LC 2/6','LC 9/7-M','LC 2/10', Low
Complexity [1]
% 'LC 5/11-C','LC 5/11-A','LC 6/14',
% 'LC 13/7-T','LC 13/7-C'
% 'Le Gall 5/3','CDF 9/7' JPEG2000 [7]
% 'V9/3' Visual [8]
% 'Lazy' Lazy wavelet
% 
Case and spaces are ignored in wavelet names, for
example, 'Sym4'
 App en d i x B 365

% m
ay also be written as 'sym 4'. Some wavelets have
multiple names,
% 
'D1', 'Sym1', and 'Spline 1.1' are aliases of the
Haar wavelet.
%
% 
WAVELET(W) displays information about wavelet W and
plots the
% primal and dual scaling and wavelet functions.
%
% 
For 2D transforms, prefix W with '2D'. For example,
'2D S+P (2,2)'
% 
specifies a 2D (tensor) transform with the S+P
(2,2) wavelet.
% 
2D transforms require that X is either MxN or MxNxP
where M and N
% are divisible by 2^L.
%
% 
WAVELET(W,L,X,EXT) specifies boundary handling EXT.
Choices are
% 'sym' Symmetric extension (same as 'wsws')
% 
'asym' Antisymmetric extension, whole-point
antisymmetry
% 'zpd' Zero-padding
% 'per' Periodic extension
% 'sp0' Constant extrapolation
%
% Various symmetric extensions are supported:
% 'wsws' Whole-point symmetry (WS) on both boundaries
% 'hshs' Half-point symmetry (HS) on both boundaries
% 'wshs' WS left boundary, HS right boundary
% 'hsws' HS left boundary, WS right boundary
%
% 
Antisymmetric boundary handling is used by default,
EXT = 'asym'.
%
% WAVELET(...,DIM) operates along dimension DIM.
%
% 
[H1,G1,H2,G2] = WAVELET(W,'filters') returns the
filters
% 
associated with wavelet transform W. Each filter is
represented
% 
by a cell array where the first cell contains an
array of
% 
coefficients and the second cell contains a scalar
of the leading
366 App en d i x B

% Z-power.
%
% 
[X,PHI1] = WAVELET(W,'phi1') returns an
approximation of the
% scaling function associated with wavelet transform W.
% 
[X,PHI1] = WAVELET(W,'phi1',N) approximates the
scaling function
% with resolution 2^-N. Similarly,
% [X,PSI1] = WAVELET(W,'psi1',...),
% [X,PHI2] = WAVELET(W,'phi2',...),
% 
and [X,PSI2] = WAVELET(W,'psi2',...) return
approximations of the
% 
wavelet function, dual scaling function, and dual
wavelet function.
%
% 
Wavelet transforms are implemented using the lifting
scheme [4].
% 
For general background on wavelets, see for example
[6].
%
%
% Examples:
% % Display information about the S+P (4,4) wavelet
% wavelet('S+P (4,4)');
%
% % Plot a wavelet decomposition
% t = linspace(0,1,256);
% X = exp(-t) + sqrt(t - 0.3).*(t > 0.3) - 0.2*​
(t > 0.6);
% 
wavelet('RSpline 3.1',3,X); % Plot the decomposition
of X
%
% % Sym4 with periodic boundaries
% 
Y = wavelet('Sym4',5,X,'per'); % Forward transform
with 5 stages
% R = wavelet('Sym4',-5,Y,'per'); % Invert 5 stages
%
% % 2D transform on an image
% t = linspace(-1,1,128); [x,y] = meshgrid(t,t);
% X = ((x+1).*(x-1) - (y+1).*(y-1)) + real(sqrt(0.4 -
x.^2 - y.^2));
% 
Y = wavelet('2D CDF 9/7',2,X); % 2D wavelet transform
% R = wavelet('2D CDF 9/7',-2,Y); % Recover X from Y
% imagesc(abs(Y).^0.2); colormap(gray); axis image;
%
 App en d i x B 367

% % Plot the Daubechies 2 scaling function

% [x,phi] = wavelet('D2','phi');
% plot(x,phi);
%
% References:
% [1] M. Adams and F. Kossentini. “Reversible Integer-

to-Integer Wavelet Transforms for Image Compression.”
IEEE Trans. on Image Proc., vol. 9, no. 6, Jun. 2000.
%
% [
2] M. Antonini, M. Barlaud, P. Mathieu, and
I. Daubechies. “Image Coding using Wavelet
Transforms.” IEEE Trans. Image Processing,vol. 1,
pp. 205-220, 1992.
%
% 
[3] R. Calderbank, I. Daubechies, W. Sweldens, and
Boon-Lock Yeo.“Lossless Image Compression using
Integer to Integer Wavelet Transforms.” ICIP IEEE
Press, vol. 1, pp. 596-599. 1997.
%
% 
[4] I. Daubechies and W. Sweldens. “Factoring Wavelet
Transforms into Lifting Steps.” 1996.
%
% 
[5] D. Le Gall and A. Tabatabai. “Subband Coding of
Digital Images Using Symmetric Short Kernel Filters
and Arithmetic Coding Techniques.” ICASSP'88,
pp.761-765, 1988.
%
% 
[6] S. Mallat. “A Wavelet Tour of Signal Processing.”
Academic Press, 1999.
%
% 
[7] M. Unser and T. Blu. “Mathematical Properties of
the JPEG2000 Wavelet Filters.” IEEE Trans. on Image
Proc., vol. 12, no. 9, Sep. 2003.
%
% 
[8] Qinghai Wang and Yulong Mo. “Choice of Wavelet
Base in JPEG2000.” Computer Engineering, vol. 30,
no. 23, Dec. 2004.
% 
Pascal Getreuer 2005-2006

if nargin < 1, error('Not enough input arguments.'); end

if ~ischar(WaveletName), error('Invalid wavelet
name.'); end

% Get a lifting scheme sequence for the specified

wavelet
368 App en d i x B

Flag1D = isempty(findstr(lower(WaveletName),'2d'));
[Seq,ScaleS,ScaleD,Family] = getwavelet(WaveletName);

if isempty(Seq)
error(['Unknown wavelet, ''',WaveletName,'''.']);
end
if nargin < 2, Level = ''; end
if ischar(Level)
[h1,g1] = seq2hg(Seq,ScaleS,ScaleD,0);
[h2,g2] = seq2hg(Seq,ScaleS,ScaleD,1);

if strcmpi(Level,'filters')
varargout = {h1,g1,h2,g2};
else
if nargin < 3, X = 6; end

switch lower(Level)
case {'phi1','phi'}
[x1,phi] = cascade(h1,g1,pow2(-X));
varargout = {x1,phi};
case {'psi1','psi'}
[x1,phi,x2,psi] = cascade(h1,g1,pow2(-X));
varargout = {x2,psi};
case 'phi2'
[x1,phi] = cascade(h2,g2,pow2(-X));
varargout = {x1,phi};
case 'psi2'
[x1,phi,x2,psi] = cascade(h2,g2,pow2(-X));
varargout = {x2,psi};
case ''
fprintf('\n%s wavelet ''%s'' ',Family,WaveletName);
if all(abs([norm(h1{1}),norm(h2{1})] - 1) < 1e-11)
fprintf('(orthogonal)\n');
else
fprintf('(biorthogonal)\n');
end

fprintf('Vanishing moments: %d analysis,%d

reconstruction\n',...
numvanish(g1{1}),numvanish(g2{1}));
fprintf('Filter lengths: %d/%d-tap\n',...
length(h1{1}),length(g1{1}));
fprintf('Implementation lifting steps: %d\n\n',...
size(Seq,1)-all([Seq{1,: }] = = 0));
 App en d i x B 369

fprintf('h1(z) =%s\n',filterstr(h1,ScaleS));
fprintf('g1(z) =%s\n',filterstr(g1,ScaleD));
fprintf('h2(z) =%s\n',filterstr(h2,1/ScaleS));
fprintf('g2(z) =%s\n\n',filterstr(g2,1/ScaleD));

[x1,phi,x2,psi] = cascade(h1,g1,pow2(-X));
subplot(2,2,1);
plot(x1,phi,'b-');
if diff(x1([1,end])) > 0, xlim(x1([1,end])); end
title('\phi_1');
subplot(2,2,3);
plot(x2,psi,'b-');
if diff(x2([1,end])) > 0, xlim(x2([1,end])); end
title('\psi_1');
[x1,phi,x2,psi] = cascade(h2,g2,pow2(-X));
subplot(2,2,2);
plot(x1,phi,'b-');
if diff(x1([1,end])) > 0, xlim(x1([1,end])); end
title('\phi_2');
subplot(2,2,4);
plot(x2,psi,'b-');
if diff(x2([1,end])) > 0, xlim(x2([1,end])); end
title('\psi_2');
set(gcf,'NextPlot','replacechildren');
otherwise
error(['Invalid parameter, ''',Level,'''.']);
end
end

return;
elseif nargin < 5
% Use antisymmetric extension by default
if nargin < 4
 if nargin < 3, error('Not enough input
arguments.'); end
Ext = 'asym';
end

Dim = min(find(size(X) ~ = 1));

if isempty(Dim), Dim = 1; end
end

if any(size(Level) ~ = 1), 
error('Invalid
decomposition level.'); end
­
370 App en d i x B

NumStages = size(Seq,1);
EvenStages = ~rem(NumStages,2);

if Flag1D % 1D Transfrom
%%% Convert N-D array to a 2-D array with dimension
Dim along the columns%%%
XSize = size(X); % Save original dimensions
N = XSize(Dim);
M = prod(XSize)/N;
Perm = [Dim: max(length(XSize),Dim),1: Dim-1];
X = double(reshape(permute(X,Perm),N,M));
if M = = 1 & nargout = = 0 & Level > 0
 % Create a figure of the wavelet decomposition
set(gcf,'NextPlot','replace');
subplot(Level+2,1,1);
plot(X);
title('Wavelet Decomposition');
axis tight; axis off;

X = feval(mfilename,WaveletName,Level,X,Ext,1);

for i = 1: Level
N2 = N;
N = 0.5*N;
subplot(Level+2,1,i+1);
a = max(abs(X(N+1: N2)))*1.1;
plot(N+1: N2,X(N+1: N2),'b-');
ylabel(['d',sprintf('_%c',num2str(i))]);
axis([N+1,N2,-a,a]);
end

subplot(Level+2,1,Level+2);
plot(X(1: N),'-');
xlabel('Coefficient Index');
ylabel('s_1');
axis tight;
set(gcf,'NextPlot','replacechildren');
varargout = {X};
return;
end

if rem(N,pow2(abs(Level))), error('Signal length

must be divisible by 2^L.'); end

if N < pow2(abs(Level)), error('Signal length

too small for transform level.'); end
 App en d i x B 3 71

if Level > = 0 % Forward transform

for i = 1: Level
Xo = X(2: 2: N,: );
 Xe = X(1: 2: N,: ) +
xfir(Seq{1,1},Seq{1,2},Xo,Ext);

for k = 3: 2: NumStages
Xo = Xo + xfir(Seq{k-1,1},Seq​
{k-1,2},Xe,Ext);
Xe = Xe + xfir(Seq{k,1},Seq{k,2},Xo,Ext);
end

if EvenStages
Xo = 
Xo + xfir(Seq{NumStages,1},​
Seq{NumStages,2},​Xe,Ext);
end

X(1: N,: ) = [Xe*ScaleS; Xo*ScaleD];

N = 0.5*N;
end
else % Inverse transform
N = N * pow2(Level);

for i = 1: -Level
N2 = 2*N;
Xe = X(1: N,: )/ScaleS;
Xo = X(N+1: N2,: )/ScaleD;
if EvenStages

Xo = Xo - xfir(Seq{NumStages,1},Seq{NumSta
ges,2},Xe,Ext);
end

for k = NumStages - EvenStages: -2: 3

Xe = Xe - xfir(Seq{k,1},Seq{k,2},Xo,Ext);
Xo = 
Xo - xfir(Seq{k-1,1},Seq{k-
1,2},Xe,Ext);
end
 X([1: 2: N2,2: 2: N2],: ) = [Xe - xfir​
(Seq{1,1},Seq{1,2},Xo,Ext); Xo];
N = N2;
end
end

X = i
permute(reshape(X,XSize(Perm)),Perm);
% Restore original array dimensions
372 App en d i x B

else % 2D Transfrom
N = size(X);
if length(N) > 3 |
any(rem(N([1,2]),pow2(abs(Level))))
error('Input size must be either MxN or MxNxP
where M and N are divisible by 2^L.');
end

if Level > = 0 % 2D Forward transform

for i = 1: Level
Xo = X(2: 2: N(1),1: N(2),: );
 Xe = X(1: 2: N(1),1: N(2),: ) +
xfir(Seq{1,1},Seq{1,2},Xo,Ext);

for k = 3: 2: NumStages
 Xo = Xo + xfir(Seq{k-1,1},Seq{k-1,2},​
Xe,Ext);
 Xe = Xe + xfir(Seq{k,1},Seq{k,2},Xo,Ext);
end

if EvenStages
 Xo = Xo + xfir(Seq{NumStages,1},Seq{NumSta
ges,2},Xe,Ext);
end

X(1: N(1),1: N(2),: ) = [Xe*ScaleS; Xo*ScaleD];

Xo = permute(X(1: N(1),2: 2: N(2),: ),[2,1,3]);

Xe = permute(X(1: N(1),1: 2: N(2),: ),[2,1,3])...
+ xfir(Seq{1,1},Seq{1,2},Xo,Ext);
for k = 3: 2: NumStages
 Xo = Xo + xfir(Seq{k-1,1},Seq{k-1,2},​
Xe,Ext);
Xe = Xe + xfir(Seq{k,1},Seq{k,2},Xo,Ext);
end
if EvenStages
 Xo = Xo + xfir(Seq{NumStages,1},Seq{NumSta
ges,2},Xe,Ext);
end

 X(1: N(1),1: N(2),: ) =

[permute(Xe,[2,1,3])*ScaleS,...
permute(Xo,[2,1,3])*ScaleD];
N = 0.5*N;
end
 App en d i x B 3 73

else % 2D Inverse transform

N = N*pow2(Level);

for i = 1: -Level
N2 = 2*N;
 Xe = permute(X(1: N2(1),1: N(2),:
),[2,1,3])/ScaleS;
 Xo = permute(X(1: N2(1),N(2)+1: N2(2),:
),[2,1,3])/ScaleD;

if EvenStages
Xo = Xo - xfir(Seq{NumStages,1},Seq{NumStages,2}
,Xe,Ext);
end

for k = NumStages - EvenStages: -2: 3

Xe = Xe - xfir(Seq{k,1},Seq{k,2},Xo,Ext);
Xo = Xo - xfir(Seq{k-1,1},Seq{k-1,2},Xe,Ext);
end
X(1: N2(1),[1: 2: N2(2),2: 2: N2(2)],: ) =...
[permute(Xe - xfir(Seq{1,1},Seq{1,2},Xo,
Ext),[2,1,3]),...
permute(Xo,[2,1,3])];
Xe = X(1: N(1),1: N2(2),: )/ScaleS;
Xo = X(N(1)+1: N2(1),1: N2(2),: )/ScaleD;

if EvenStages
 Xo = Xo - xfir(Seq{NumStages,1},Seq{NumStages,2}
,Xe,Ext);
end
for k = NumStages - EvenStages: -2: 3
Xe = Xe - xfir(Seq{k,1},Seq{k,2},Xo,Ext);
 Xo = Xo - xfir(Seq{k-1,1},Seq{k-
1,2},Xe,Ext);
end
 X([1: 2: N2(1),2: 2: N2(1)],1: N2(2),: )
=...
[Xe - xfir(Seq{1,1},Seq{1,2},Xo,Ext); Xo];
N = N2;
end
end
end

varargout{1} = X;
return;
3 74 App en d i x B

function [Seq,ScaleS,ScaleD,Family] =
getwavelet(WaveletName)
%GETWAVELET Get wavelet lifting scheme sequence.
% Pascal Getreuer 2005-2006

WaveletName = strrep(WaveletName,'bior','spline');
ScaleS = 1/sqrt(2);
ScaleD = 1/sqrt(2);
Family = 'Spline';

switch strrep(strrep(lower(WaveletName),'2d',''),' ','')

case {'haar','d1','db1','sym1','spline1.1','rspline1.1'}
Seq = {1,0; -0.5,0};
ScaleD = -sqrt(2);
Family = 'Haar';
case {'d2','db2','sym2'}
 Seq = {sqrt(3) ,0; [-sqrt(3),2-sqrt(3)]/4,0;
-1,1};
ScaleS = (sqrt(3)-1)/sqrt(2);
ScaleD = (sqrt(3)+1)/sqrt(2);
Family = 'Daubechies';
case {'d3','db3','sym3'}
 Seq = {2.4254972439123361,0; [-0.352387657680182
3,0.0793394561587384] ,0;
[0.5614149091879961,-2.8953474543648969],2;​
-0.0197505292372931,-2};
ScaleS = 0.4318799914853075;
ScaleD = 2.3154580432421348;
Family = 'Daubechies';
case {'d4','db4'}
Seq = {0.3222758879971411,-1;​[0.3001422587485443
,1.1171236051605939],1;
[-0.1176480867984784,0.0188083527262439],-1;
[-0.6364282711906594,-2.1318167127552199],1;
[0.0247912381571950,-
0.1400392377326117,0.4690834789110281],2};
ScaleS = 1.3621667200737697;
ScaleD = 0.7341245276832514;
Family = 'Daubechies';
case {'d5','db5'}
 Seq = {0.2651451428113514,-1;​
[-0.2477292913288009,-0.9940591341382633],1;
[-0.2132742982207803,0.5341246460905558],1;
[0.7168557197126235,-0.2247352231444452],-1;
 App en d i x B 375

[-0.0121321866213973,
0.0775533344610336],3;0.035764924629411,-3};
ScaleS = 1.3101844387211246;
ScaleD = 0.7632513182465389;
Family = 'Daubechies';
case {'d6','db6'}
Seq = {4.4344683000391223,0;​[-0.214593449940913,
0.0633131925095066],0;
[4.4931131753641633,-9.970015617571832],2;
[-0.0574139367993266,0.0236634936395882],-2;
[0.6787843541162683,-2.3564970162896977],4;
[-0.0071835631074942,0.0009911655293238],-4;​
-0.0941066741175849,5};
 ScaleS = 0.3203624223883869;
ScaleD = 3.1214647228121661;
Family = 'Daubechies';
case 'sym4'
 Seq = {-0.3911469419700402,0; [0.339243991864945
1,0.1243902829333865],0;
[-0.1620314520393038,1.4195148522334731],1;
-[0.1459830772565225,
0.4312834159749964],1;1.049255198049293,-1};
ScaleS = 0.6366587855802818;
ScaleD = 1.5707000714496564;
Family = 'Symlet';
case 'sym5'
 Seq = {0.9259329171294208,0; -[0.131923027028234
1,0.4985231842281166],1;
[1.452118924420613,0.4293261204657586],0;
[-0.2804023843755281,0.0948300395515551],0;
-[0.7680659387165244,1.9589167118877153],1;​
0.1726400850543451,0};
ScaleS = 0.4914339446751972;
ScaleD = 2.0348614718930915;
Family = 'Symlet';
case 'sym6'
Seq = {-0.2266091476053614,0;
[0.2155407618197651,-1.2670686037583443],0;
[-4.2551584226048398,0.5047757263881194],2;
[0.2331599353469357 =,0.0447459687134724],-2;
[6.6244572505007815,-18.389000853969371],4;
[-0.0567684937266291,0.1443950619899142],-4;​
-5.5119344180654508,5};
ScaleS = -0.5985483742581210;
ScaleD = -1.6707087396895259;
376 App en d i x B

Family = 'Symlet';
case 'coif1'
 Seq = {-4.6457513110481772,0;[0.205718913884,
0.1171567416519999],0;
[0.6076252184992341,-7.468626966435207],2;​
-0.0728756555332089,-2};
ScaleS = -0.5818609561112537;
ScaleD = -1.7186236496830642;
Family = 'Coiflet';
case 'coif2'
 Seq = {-2.5303036209828274,0;
[0.3418203790296641,-0.2401406244344829],0;
[15.268378737252995,3.1631993897610227],2;
[-0.0646171619180252,0.005717132970962],-2;
[13.59117256930759,-63.95104824798802],4;
[-0.0018667030862775,0.0005087264425263],-4;​
-3.7930423341992774,5};
ScaleS = 0.1076673102965570;
ScaleD = 9.2878701738310099;
Family = 'Coiflet';
case 'bcoif1'
 Seq = {0,0; -[1,1]/5,1;[5,5]/14,0;
-[21,21]/100,1};
ScaleS = sqrt(2)*7/10;
ScaleD = sqrt(2)*5/7;
Family = 'Nearly orthonormal Coiflet-like';
case {'lazy','spline0.0','rspline0.0','d0'}
Seq = {0,0};
ScaleS = 1;
ScaleD = 1;
Family = 'Lazy';
case {'spline0.1','rspline0.1'}
Seq = {1,-1};
ScaleD = 1;
case {'spline0.2','rspline0.2'}
Seq = {[1,1]/2,0};
ScaleD = 1;
case {'spline0.3','rspline0.3'}
Seq = {[-1,6,3]/8,1};
ScaleD = 1;
case {'spline0.4','rspline0.4'}
Seq = {[-1,9,9,-1]/16,1};
ScaleD = 1;
case {'spline0.5','rspline0.5'}
Seq = {[3,-20,90,60,-5]/128,2};
 App en d i x B 377

ScaleD = 1;
case {'spline0.6','rspline0.6'}
Seq = {[3,-25,150,150,-25,3]/256,2};
ScaleD = 1;
case {'spline0.7','rspline0.7'}
Seq = {[-5,42,-175,700,525,-70,7]/1024,3};
ScaleD = 1;
case {'spline0.8','rspline0.8'}
 Seq = {[-5,49,-245,1225,1225,-245,49,-
5]/2048,3};
ScaleD = 1;
case {'spline1.0','rspline1.0'}
Seq = {0,0; -1,0};
ScaleS = sqrt(2);
ScaleD = -1/sqrt(2);
case {'spline1.3','rspline1.3'}
Seq = {0,0; -1,0; [-1,8,1]/16,1};
ScaleS = sqrt(2);
ScaleD = -1/sqrt(2);
case {'spline1.5','rspline1.5'}
Seq = {0,0; -1,0; [3,-22,128,22,-3]/256,2};
ScaleS = sqrt(2);
ScaleD = -1/sqrt(2);
case {'spline1.7','rspline1.7'}
 Seq = {0,0; -1,0;
[-5,44,-201,1024,201,-44,5]/2048,3};
ScaleS = sqrt(2);
ScaleD = -1/sqrt(2);
case {'spline2.0','rspline2.0'}
Seq = {0,0; -[1,1]/2,1};
ScaleS = sqrt(2);
ScaleD = 1;
case {'spline2.1','rspline2.1'}
Seq = {0,0; -[1,1]/2,1;0.5,0};
ScaleS = sqrt(2);
case {'spline2.2','rspline2.2','cdf5/3','legall5/3','s
+p(2,2)','lc5/3'}
Seq = {0,0; -[1,1]/2,1;[1,1]/4,0};
ScaleS = sqrt(2);
case {'spline2.4','rspline2.4'}
Seq = {0,0; -[1,1]/2,1;[-3,19,19,-3]/64,1};
ScaleS = sqrt(2);
case {'spline2.6','rspline2.6'}
Seq = 
{0,0; -[1,1]/2,1;[5,-39,162,162,-
39,5]/512,2};
3 78 App en d i x B

ScaleS = sqrt(2);
case {'spline2.8','rspline2.8'}
Seq = 
{0,0; -[1,1]/2,1;[-35,335,-1563,​
5359,5359,-1563,335,-35]/16384,3};
ScaleS = sqrt(2);
case {'spline3.0','rspline3.0'}
Seq = {-1/3,-1;-[3,9]/8,1};
ScaleS = 3/sqrt(2);
ScaleD = 2/3;
case {'spline3.1','rspline3.1'}
Seq = {-1/3,-1;-[3,9]/8,1;4/9,0};
ScaleS = 3/sqrt(2);
ScaleD = -2/3;
case {'spline3.3','rspline3.3'}
Seq = {-1/3,-1;-[3,9]/8,1;[-3,16,3]/36,1};
ScaleS = 3/sqrt(2);
ScaleD = -2/3;
case {'spline3.5','rspline3.5'}
Seq = 
{-1/3,-1;-[3,9]/8,1;[5,-34,128,34,-5]​
/288,2};
ScaleS = 3/sqrt(2);
ScaleD = -2/3;
case {'spline3.7','rspline3.7'}
Seq = 
{-1/3,-1;-[3,9]/8,1;[-35,300,-1263,​
4096,1263,-300,35]/9216,3};
ScaleS = 3/sqrt(2);
ScaleD = -2/3;
case {'spline4.0','rspline4.0'}
Seq = {-[1,1]/4,0; -[1,1],1};
ScaleS = 4/sqrt(2);
ScaleD = 1/sqrt(2);
ScaleS = 1; ScaleD = 1;
case {'spline4.1','rspline4.1'}
Seq = {-[1,1]/4,0; -[1,1],1;6/16,0};
ScaleS = 4/sqrt(2);
ScaleD = 1/2;
case {'spline4.2','rspline4.2'}
Seq = {-[1,1]/4,0; -[1,1],1;[3,3]/16,0};
ScaleS = 4/sqrt(2);
ScaleD = 1/2;
case {'spline4.4','rspline4.4'}
Seq = 
{-[1,1]/4,0; -[1,1],1;[-5,29,29,-5]​
/128,1};
ScaleS = 4/sqrt(2);
ScaleD = 1/2;
 App en d i x B 3 79

case {'spline4.6','rspline4.6'}
Seq = 
{-[1,1]/4,0; -[1,1],1;[35,-265,998,998,-
265,35]/4096,2};
ScaleS = 4/sqrt(2);
ScaleD = 1/2;
case {'spline4.8','rspline4.8'}
Seq = 
{-[1,1]/4,0; -[1,1],1;[-63,595,-2687,​
8299,8299,-2687,595,-63]/32768,3};
ScaleS = 4/sqrt(2);
ScaleD = 1/2;
case {'spline5.0','rspline5.0'}
Seq = 
{0,0; -1/5,0; -[5,15]/24,0; -[9,15]/10,1};
ScaleS = 3*sqrt(2);
ScaleD = sqrt(2)/6;
case {'spline5.1','rspline5.1'}
Seq = 
{0,0; -1/5,0; -[5,15]/24,0;
-[9,15]/10,1;1/3,0};
ScaleS = 3*sqrt(2);
ScaleD = sqrt(2)/6;
case {'spline5.3','rspline5.3'}
Seq = 
{0,0; -1/5,0; -[5,15]/24,0;
-[9,15]/10,1;[-5,24,5]/72,1};
ScaleS = 3*sqrt(2);
ScaleD = sqrt(2)/6;
case {'spline5.5','rspline5.5'}
Seq = 
{0,0; -1/5,0; -[5,15]/24,0;
-[9,15]/10,1;[35,-230,768,230,-35]​
/2304,2};
ScaleS = 3*sqrt(2);
ScaleD = sqrt(2)/6;
case {'cdf9/7'}
Seq = 
{0,0; [1,1]*-1.5861343420693648,1;[1,1]​
*-0.0529801185718856,0;
[1,1]*0.8829110755411875,1;[1,1]*0.4435068520511
142,0};
ScaleS = 1.1496043988602418;
ScaleD = 1/ScaleS;
Family = 'Cohen-Daubechies-Feauveau';
case 'v9/3'
Seq = {0,0; [-1,-1]/2,1;[1,19,19,1]/80,1};
ScaleS = sqrt(2);
Family = 'HSV design';
case {'s+p(4,2)','lc9/7-m'}
Seq = {0,0; [1,-9,-9,1]/16,2;[1,1]/4,0};
ScaleS = sqrt(2);
380 App en d i x B

Family = 'S+P';
case 's+p(6,2)'
Seq = {0,0; [-3,25,-150,-150,25,-3]​
/256,3;[1,1]/4,0};
ScaleS = sqrt(2);
Family = 'S+P';
case {'s+p(4,4)','lc13/7-t'}
Seq = {0,0; [1,-9,-9,1]/16,2;[-1,9,9,-1]/32,1};
ScaleS = sqrt(2);
Family = 'S+P';
case {'s+p(2+2,2)','lc5/11-c'}
Seq = {0,0; [-1,-1]/2,1;[1,1]/4,0; -[-1,1,1,-1]​
/16,2};
ScaleS = sqrt(2);
Family = 'S+P';
case 'tt'
Seq = {1,0; [3,-22,-128,22,-3]/256,2};
ScaleD = sqrt(2);
Family = 'Le Gall-Tabatabai polynomial';
case 'lc2/6'
Seq = {0,0; -1,0; 1/2,0; [-1,0,1]/4,1};
ScaleS = sqrt(2);
ScaleD = -1/sqrt(2);
Family = 'Reverse spline';
case 'lc2/10'
Seq = {0,0; -1,0; 1/2,0; [3,-22,0,22,-3]/64,2};
ScaleS = sqrt(2);
ScaleD = -1/sqrt(2);
Family = 'Reverse spline';
case 'lc5/11-a'
Seq = {0,0; -[1,1]/2,1;[1,1]/4,0; [1,-1,-1,1]​
/32,2};
ScaleS = sqrt(2);
ScaleD = -1/sqrt(2);
Family = 'Low complexity';
case 'lc6/14'
Seq = {0,0; -1,0; [-1,8,1]/16,1;[1,-6,0,6,-1]​
/16,2};
ScaleS = sqrt(2);
ScaleD = -1/sqrt(2);
Family = 'Low complexity';
case 'lc13/7-c'
Seq = {0,0; [1,-9,-9,1]/16,2;[-1,5,5,-1]/16,1};
ScaleS = sqrt(2);
ScaleD = -1/sqrt(2);
 App en d i x B 3 81

Family = 'Low complexity';

otherwise
Seq = {};
return;
end

if ~isempty(findstr(lower(WaveletName),'rspline'))
[Seq,ScaleS,ScaleD] = seqdual(Seq,ScaleS,ScaleD);
Family = 'Reverse spline';
end

return;

function [Seq,ScaleS,ScaleD] =
seqdual(Seq,ScaleS,ScaleD)
% Dual of a lifting sequence

L = size(Seq,1);

for k = 1: L
% f'(z) = -f(z^-1)
Seq{k,2} = -(Seq{k,2} - length(Seq{k,1}) + 1);
Seq{k,1} = -fliplr(Seq{k,1});
end

if all(Seq{1,1} = = 0)
Seq = reshape({Seq{2: end,: }},L-1,2);
else
 [Seq{1: L+1,: }] = deal(0,Seq{1: L,1},0,Seq{1:
L,2});
end

ScaleS = 1/ScaleS;
ScaleD = 1/ScaleD;
return;

function [h,g] = seq2hg(Seq,ScaleS,ScaleD,Dual)

% Find wavelet filters from lifting sequence
if Dual, [Seq,ScaleS,ScaleD] =
seqdual(Seq,ScaleS,ScaleD); end
if rem(size(Seq,1),2), [Seq{size(Seq,1)+1,: }] =
deal(0,0); end

h = {1,0};
g = {1,1};
382 App en d i x B

for k = 1: 2: size(Seq,1)
h = lp_lift(h,g,{Seq{k,: }});
g = lp_lift(g,h,{Seq{k+1,: }});
end

h = {ScaleS*h{1},h{2}};
g = {ScaleD*g{1},g{2}};

if Dual
h{2} = -(h{2} - length(h{1}) + 1);
h{1} = fliplr(h{1});

g{2} = -(g{2} - length(g{1}) + 1);

g{1} = fliplr(g{1});
end

return;

function a = lp_lift(a,b,c)
% a(z) = a(z) + b(z) c(z^2)

d = zeros(1,length(c{1})*2-1);
d(1: 2: end) = c{1};
d = conv(b{1},d);
z = b{2}+c{2}*2;
zmax = max(a{2},z);
f = [zeros(1,zmax-
a{2}),a{1},zeros(1,a{2} - length(a{1}) - z +
length(d))];
i = zmax-z + (1: length(d));
f(i) = f(i) + d;

if all(abs(f) < 1e-12)

a = {0,0};
else
i = find(abs(f)/max(abs(f)) > 1e-10);
i1 = min(i);
a = {f(i1: max(i)),zmax-i1+1};
end
return;

function X = xfir(B,Z,X,Ext)
%XFIR Noncausal FIR filtering with boundary handling.
% Y = XFIR(B,Z,X,EXT) filters X with FIR filter B with
leading
 App en d i x B 383

% delay -Z along the columns of X. EXT specifies the

boundary
% handling. Special handling is done for one and two-
tap filters.

% Pascal Getreuer 2005-2006

N = size(X);

% Special handling for short filters

if length(B) = = 1 & Z = = 0
if B = = 0
X = zeros(size(X));
elseif B ~ = 1
X = B*X;
end
return;
end

% Compute the number of samples to add to each end of

the signal
pl = max(length(B)-1-Z,0); % Padding on the left end
pr = max(Z,0); % Padding on the right end

switch lower(Ext)
case {'sym','wsws'} % Symmetric extension, WSWS
if all([pl,pr] < N(1))
X = 
filter(B,1,X([pl+1: -1: 2,1:
N(1),N(1)-1: -1: N(1)-pr],:,: ),[],1);
X = X(Z+pl+1: Z+pl+N(1),:,: );
return;
else
i = [1: N(1),N(1)-1: -1: 2];
Ns = 2*N(1) - 2 + (N(1) = = 1);
i = 
i([rem(pl*(Ns-1): pl*Ns-1,Ns)+1,1:
N(1),rem(N(1): N(1)+pr-1,Ns)+1]);
end
case {'symh','hshs'} % Symmetric extension, HSHS
if all([pl,pr] < N(1))
i = 
[pl: -1: 1,1: N(1),N(1): -1:
N(1)-pr+1];
else
i = [1: N(1),N(1): -1: 1];
Ns = 2*N(1);
384 App en d i x B

i = 
i([rem(pl*(Ns-1): pl*Ns-1,Ns)+1,1:
N(1),rem(N(1): N(1)+pr-1,Ns)+1]);
end
case 'wshs' % Symmetric extension, WSHS
if all([pl,pr] < N(1))
i = 
[pl+1: -1: 2,1: N(1),N(1): -1:
N(1)-pr+1];
else
i = [1: N(1),N(1): -1: 2];
Ns = 2*N(1) - 1;
i = 
i([rem(pl*(Ns-1): pl*Ns-1,Ns)+1,1:
N(1),rem(N(1): N(1)+pr-1,Ns)+1]);
end
case 'hsws' % Symmetric extension, HSWS
if all([pl,pr] < N(1))
 i = [pl: -1: 1,1: N(1),N(1)-1: -1:
N(1)-pr];
else
i = [1: N(1),N(1)-1: -1: 1];
Ns = 2*N(1) - 1;
 i = i([rem(pl*(Ns-1): pl*Ns-1,Ns)+1,1:
N(1),rem(N(1): N(1)+pr-1,Ns)+1]);
end
case 'zpd'
Ml = N; Ml(1) = pl;
Mr = N; Mr(1) = pr;
X = filter(B,1,[zeros(Ml);X;zeros(Mr)],[],1);
X = X(Z+pl+1: Z+pl+N(1),:,: );
return;
case 'per' % Periodic
i = [rem(pl*(N(1)-1): pl*N(1)-1,N(1))+1,1:
N(1),rem(0: pr-1,N(1))+1];
case 'sp0' % Constant extrapolation
i = [ones(1,pl),1: N(1),N(1)+zeros(1,pr)];
case 'asym' % Asymmetric extension
i1 = [ones(1,pl),1: N(1),N(1)+zeros(1,pr)];

if all([pl,pr] < N(1))

i2 = 
[pl+1: -1: 2,1: N(1),N(1)-1: -1:
N(1)-pr];
else
i2 = [1: N(1),N(1)-1: -1: 2];
Ns = 2*N(1) - 2 + (N(1) = = 1);
i2 = 
i2([rem(pl*(Ns-1): pl*Ns-1,Ns)+1,1:
N(1),rem(N(1): N(1)+pr-1,Ns)+1]);
 App en d i x B 385

end

X = 
filter(B,1,2*X(i1,:,: ) - X(i2,:,: ),[],1);
X = 
X(Z+pl+1: Z+pl+N(1),:,: );
return;
otherwise
 error(['Unknown boundary handling,
''',Ext,'''.']);
end

X = filter(B,1,X(i,:,: ),[],1);
X = X(Z+pl+1: Z+pl+N(1),:,: );
return;

function [x1,phi,x2,psi] = cascade(h,g,dx)

% Wavelet cascade algorithm

c = h{1}*2/sum(h{1});
x = 0: dx: length(c) - 1;
x1 = x - h{2};
phi0 = 1 - abs(linspace(-1,1,length(x))).';

ii = []; jj = []; s = [];

for k = 1: length(c)
xk = 2*x - (k-1);
i = find(xk > = 0 & xk < = length(c) - 1);
ii = [ii,i];
jj = [jj,floor(xk(i)/dx)+1];
s = [s,c(k)+zeros(size(i))];
end

% Construct a sparse linear operator that iterates the

dilation equation

Dilation = sparse(ii,jj,s,length(x),length(x));

for N = 1: 30
phi = Dilation*phi0;
if norm(phi - phi0,inf) < 1e-5, break; end
phi0 = phi;
end

if norm(phi) = = 0
phi = o
nes(size(phi))*sqrt(2); % Special case
for Haar scaling function
386 App en d i x B

else
phi = phi/(norm(phi)*sqrt(dx)); % Rescale result
end

if nargout > 2
phi2 = phi(1: 2: end); % phi2 is approximately
phi(2x)
if length(c) = = 2
L = length(phi2);
else
L = ceil(0.5/dx);
end

% Construct psi from translates of phi2

c = g{1};
psi = zeros(length(phi2)+L*(length(c)-1),1);
x2 = (0: length(psi)-1)*dx - g{2} - 0*h{2}/2;

for k = 1: length(c)
i = (1: length(phi2)) + L*(k-1);
psi(i) = psi(i) + c(k)*phi2;
end
end
return;

function s = filterstr(a,K)
% Convert a filter to a string
[n,d] = rat(K/sqrt(2));
if d < 50
a{1} = a{1}/sqrt(2); % Scale filter by sqrt(2)
s = '(';
else
s = '';
end

Scale = [pow2(1: 15),10,20,160,280,inf];

for i = 1: length(Scale)
 if norm(round(a{1}*Scale(i))/Scale(i) -
a{1},inf) < 1e-9
 a{1} = a{1}*Scale(i); % Scale filter by a
power of 2 or 160
s = '(';
break;
end
end
 App en d i x B 387

z = a{2};
LineOff = 0;

for k = 1: length(a{1})
v = a{1}(k);
if v ~ = 0 % Only display nonzero coefficients
if k > 1
s2 = [' ',char(44-sign(v)),' '];
v = abs(v);
else
s2 = '';
end
s2 = sprintf('%s%g',s2,v);
if z = = 1
s2 = sprintf('%s z',s2);
elseif z ~ = 0
s2 = sprintf('%s z^%d',s2,z);
end
 if length(s) + length(s2) > 72 + LineOff % Wrap
long lines
s2 = [char(10),' ',s2];
LineOff = length(s);
end
s = [s,s2];
end
z = z - 1;
end

if s(1) = = '('
s = [s,')'];
if d < 50, s = [s,' sqrt(2)']; end

if i < length(Scale)
s = sprintf('%s/%d',s,Scale(i));
end
end

return;

function N = numvanish(g)
% Determine the number of vanishing moments from
highpass filter g(z)

for N = 0: length(g)-1 % Count the number of roots at

z = 1
388 App en d i x B

[g,r] = deconv(g,[1,-1]);
if norm(r,inf) > 1e-7, break; end
end
return;

function [iout,qout] = delay(idata,qdata,nsamp,idel)

iout = zeros(1,nsamp);
qout = zeros(1,nsamp);
if idel ~ = 0
iout(1: idel) = zeros(1,idel);
qout(1: idel) = zeros(1,idel);
end
iout(idel+1: nsamp) = idata(1: nsamp-idel);
qout(idel+1: nsamp) = qdata(1: nsamp-idel);

function [iout,qout] = comb(idata,qdata,attn)

iout = randn(1,length(idata)).*attn;
qout = randn(1,length(qdata)).*attn;
iout = iout+idata(1: length(idata));
qout = qout+qdata(1: length(qdata));

function [iout,qout] = 
giins(idata,qdata,fftlen,gilen
,nd);
idata1 = reshape(idata,fftlen,nd);
qdata1 = reshape(qdata,fftlen,nd);
idata2 = [idata1(fftlen-gilen+1: fftlen,: );idata1];
qdata2 = [qdata1(fftlen-gilen+1: fftlen,: );qdata1];
iout = reshape(idata2,1,(fftlen+gilen)*nd);
qout = reshape(qdata2,1,(fftlen+gilen)*nd);

function [iout,qout] = 
girem(idata,qdata,fftlen2,gilen
,nd);
idata2 = reshape(idata,fftlen2,nd);
qdata2 = reshape(qdata,fftlen2,nd);
iout = idata2(gilen+1: fftlen2,: );
qout = qdata2(gilen+1: fftlen2,: );

function [iout,qout] = 
giins1(idata,qdata,fftlen,​
gilen,nd);
idata1 = reshape(idata,fftlen,nd);
qdata1 = reshape(qdata,fftlen,nd);
gg = gilen/2;
idata2 = [
zeros(gg,nd);idata1(fftlen-gg+1: fftlen,: )​
;idata1];
 App en d i x B 389

qdata2 = [
zeros(gg,nd);qdata1(fftlen-gg+1:
fftlen,: )​;qdata1];
iout = reshape(idata2,1,(fftlen+gilen)*nd);
qout = reshape(qdata2,1,(fftlen+gilen)*nd);

function[iout,qout,ramp,rcos,rsin] = sefade(idata,qdat
a,itau,dlvl,th,n0,itn,n1,nsamp,tstp,fd,flat)
iout = zeros(1,nsamp);;
qout = zeros(1,nsamp);;
total_attn = sum(10.^(-1 .0. *dlvl./10.0));
for k = 1: n1
atts = 10.^(-0.05.*dlvl(k));
theta = th(k).*pi./180.0;
[itmp,qtmp] = delay(idata,qdata,nsamp,itau(k));
 [itmp3,qtmp3,ramp,rcos,rsin] = fade(itmp,qtmp,ns
amp,tstp,fd,n0(k),itn(k),flat);
iout = iout+atts.*itmp3./sqrt(total_attn);
qout = qout+atts.*qtmp3./sqrt(total_attn);
end

function [iout,qout] = 
girem1(idata,qdata,fftlen2,gile
n,nd);
idata2 = reshape(idata,fftlen2,nd);
qdata2 = reshape(qdata,fftlen2,nd);
iout = idata2(gilen+1: fftlen2,: );
qout = qdata2(gilen+1: fftlen2,: );

function [iout,qout] = 
giins1(idata,qdata,fftlen,gilen
,nd);
idata1 = reshape(idata,fftlen,nd);
qdata1 = reshape(qdata,fftlen,nd);
idata2 = [zeros(gilen,nd);idata1];
qdata2 = [zeros(gilen,nd);qdata1];
iout = reshape(idata2,1,(fftlen+gilen)*nd);
qout = reshape(qdata2,1,(fftlen+gilen)*nd);
Digital & Wireless Communications

Presenting encryption algorithms with varying characteristics, Image Encryption:

A Communication Perspective examines image encryption algorithms for the
purpose of secure wireless communication. It considers two directions for image
encryption: permutation-based approaches and substitution-based approaches.

Covering the spectrum of image encryption principles and techniques, the book
compares image encryption with permutation- and diffusion-based approaches. It
explores number theory-based encryption algorithms such as the Data Encryption
Standard, the Advanced Encryption Standard, and the RC6 algorithms. It not only
details the strength of various encryption algorithms, but also describes their
ability to work within the limitations of wireless communication systems.

Since some ciphers were not designed for image encryption, the book explains
how to modify these ciphers to work for image encryption. It also provides
instruction on how to search for other approaches suitable for this task. To
make this work comprehensive, the authors explore communication concepts
concentrating on the orthogonal frequency division multiplexing (OFDM) system
and present a simplified model for the OFDM communication system with its
different implementations.

Complete with simulation experiments and MATLAB® codes for most of the
simulation experiments, this book will help you gain the understanding required
to select the encryption method that best fulfills your application requirements.

K16760
6000 Broken Sound Parkway, NW ISBN: 978-1-4665-7698-8
Suite 300, Boca Raton, FL 33487 90000
711 Third Avenue
New York, NY 10017
an informa business 2 Park Square, Milton Park
www.crcpress.com Abingdon, Oxon OX14 4RN, UK
9 781466 576988
w w w.crcpress.com