DSP for MATLAB and LabVIEW Fundamentals of

discrete frequency transforms Forester W. Isen

pdf download

https://ebookgate.com/product/dsp-for-matlab-and-labview-
fundamentals-of-discrete-frequency-transforms-forester-w-isen/

Get Instant Ebook Downloads – Browse at https://ebookgate.com

Instant digital products (PDF, ePub, MOBI) available
Download now and explore formats that suit you...

DSP for MATLAB and LabVIEW I Fundamentals of Discrete

Signal Processing Synthesis Lectures on Signal Processing
Isen
https://ebookgate.com/product/dsp-for-matlab-and-labview-i-
fundamentals-of-discrete-signal-processing-synthesis-lectures-on-
signal-processing-isen/
ebookgate.com

Hilbert Transforms Volume 1 Encyclopedia of Mathematics

and its Applications 1st Edition Frederick W. King

https://ebookgate.com/product/hilbert-transforms-
volume-1-encyclopedia-of-mathematics-and-its-applications-1st-edition-
frederick-w-king/
ebookgate.com

Radio Frequency Circuit Design 1st Edition W. Alan Davis

https://ebookgate.com/product/radio-frequency-circuit-design-1st-
edition-w-alan-davis/

ebookgate.com

Discrete Fourier analysis 1st Edition M. W. Wong (Auth.)

https://ebookgate.com/product/discrete-fourier-analysis-1st-edition-m-
w-wong-auth/

ebookgate.com
A Frequency Dictionary of Dutch Core Vocabulary for
Learners Routledge Frequency Dictionaries 1st Edition
Carole Tiberius
https://ebookgate.com/product/a-frequency-dictionary-of-dutch-core-
vocabulary-for-learners-routledge-frequency-dictionaries-1st-edition-
carole-tiberius/
ebookgate.com

Labview for Everyone Graphical Programming Made Easy and

Fun 3rd Edition Edition Travis

https://ebookgate.com/product/labview-for-everyone-graphical-
programming-made-easy-and-fun-3rd-edition-edition-travis/

ebookgate.com

Radio Frequency Integrated Circuit Design 2nd Edition John

M. W. Rogers

https://ebookgate.com/product/radio-frequency-integrated-circuit-
design-2nd-edition-john-m-w-rogers/

ebookgate.com

Instructor Solution Manual Fundamentals of Digital Signal

Processing Using MATLAB 2nd Edition Robert J. Schilling

https://ebookgate.com/product/instructor-solution-manual-fundamentals-
of-digital-signal-processing-using-matlab-2nd-edition-robert-j-
schilling/
ebookgate.com

Models for discrete data Daniel Zelterman

https://ebookgate.com/product/models-for-discrete-data-daniel-
zelterman/

ebookgate.com
DSP for
MATLAB™ and LabVIEW™
Volume II: Discrete Frequency
Transforms
 SYNTHESIS LECTURES
ON SIGNAL
PROCESSING
Editor
José Moura, Carnegie Mellon University

DSP for MATLAB™ and LabVIEW™ Volume II: Discrete Frequency Transforms
Forester W. Isen
2008

DSP for MATLAB™ and LabVIEW™

Volume I: Fundamentals of Discrete Signal Processing
Forester W. Isen
2008

The Theory of Linear Prediction

P. P. Vaidyanathan
2007

Nonlinear Source Separation

Luis B. Almeida
2006

Spectral Analysis of Signals

Yanwei Wang, Jian Li, and Petre Stoica
2006
Copyright © 2008 by Morgan & Claypool

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in
any form or by any means—electronic, mechanical, photocopy, recording, or any other except for brief quotations in
printed reviews, without the prior permission of the publisher.

DSP for MATLAB™ and LabVIEW™ Volume II: Discrete Frequency Transforms
Forester W. Isen
www.morganclaypool.com

ISBN: 9781598298932 paperback

ISBN: 9781598298949 ebook

DOI 10.2200/S000162ED1V01Y200811SPR005

A Publication in the Morgan & Claypool Publishers series

SYNTHESIS LECTURES ON SIGNAL PROCESSING

Lecture #5
Series Editor: José Moura, Carnegie Mellon University

Series ISSN
Synthesis Lectures on Signal Processing
Print 1932-1236 Electronic 1932-1694
DSP for
MATLAB™ and LabVIEW™
Volume II: Discrete Frequency
Transforms

Forester W. Isen

SYNTHESIS LECTURES ON SIGNAL PROCESSING #5

M
&C Morgan & cLaypool publishers
ABSTRACT
This book is Volume II of the series DSP for MATLAB ™ and LabVIEW ™ . This volume provides detailed
coverage of discrete frequency transforms, including a brief overview of common frequency transforms,
both discrete and continuous, followed by detailed treatments of the Discrete Time Fourier Transform
(DTFT), the z-Transform (including definition and properties, the inverse z-transform, frequency re-
sponse via z-transform, and alternate filter realization topologies including Direct Form, Direct Form
Transposed, Cascade Form, Parallel Form, and Lattice Form), and the Discrete Fourier Transform (DFT)
(including Discrete Fourier Series, the DFT-IDFT pair, DFT of common signals, bin width, sampling
duration, and sample rate, the FFT, the Goertzel Algorithm, Linear, Periodic, and Circular convolu-
tion, DFT Leakage, and computation of the Inverse DFT). The entire series consists of four volumes
that collectively cover basic digital signal processing in a practical and accessible manner, but which
nonetheless include all essential foundation mathematics. As the series title implies, the scripts (of which
there are more than 200) described in the text and supplied in code form (available via the internet at
http://www.morganclaypool.com/page/isen) will run on both MATLAB and LabVIEW. The
text for all volumes contains many examples, and many useful computational scripts, augmented by
demonstration scripts and LabVIEW Virtual Instruments (VIs) that can be run to illustrate various
signal processing concepts graphically on the user’s computer screen. Volume I consists of four chapters
that collectively set forth a brief overview of the field of digital signal processing, useful signals and
concepts (including convolution, recursion, difference equations, LTI systems, etc), conversion from the
continuous to discrete domain and back (i.e., analog-to-digital and digital-to-analog conversion), alias-
ing, the Nyquist rate, normalized frequency, sample rate conversion, and Mu-law compression, and signal
processing principles including correlation, the correlation sequence, the Real DFT, correlation by con-
volution, matched filtering, simple FIR filters, and simple IIR filters. Chapter 4 of Volume I, in particular,
provides an intuitive or “first principle” understanding of how digital filtering and frequency transforms
work, preparing the reader for the present volume (Volume II). Volume III of the series covers digital
filter design (FIR design using Windowing, Frequency Sampling, and Optimum Equiripple techniques,
and Classical IIR design) and Volume IV, the culmination of the series, is an introductory treatment of
LMS Adaptive Filtering and applications.

KEYWORDS
Higher-Level Terms:
MATLAB, LabVIEW, MathScript, DSP (Digital Signal Processing), Discrete Time
Fourier Transform (DTFT), z-Transform, Discrete Fourier Transform (DFT), Fast Fourier
Transform (FFT), Goertzel Algorithm, Discrete Fourier Series (DFS), Frequency Domain,
Discrete Frequency Transform
Lower-Level Terms:
FIR, IIR, Direct Form, Direct Form Transposed, Parallel Form, Cascade Form, Lattice
Form, Decimation-in-time (DIT)
This volume is dedicated to the memory of the following
Douglas Hunter (1941-1963)
Diane Satterwhite (1949-1961)
John G. Elsberry (1949-1971)
Amelia Megan Au (1967-2007)
 ix

Contents
SYNTHESIS LECTURES ON SIGNAL PROCESSING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii

Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix

Preface to Volume II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .xv

0.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv
0.2 The Four Volumes of the Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv
0.3 Origin and Evolution of the Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvi

1 The Discrete Time Fourier Transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 In the Previous Volume 1
1.1.2 In this Volume 1
1.1.3 In this Chapter 1
1.2 Software for Use with this Book . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.3 Introduction to Transform Families . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3.1 Fourier Family (Constant Unity-Magnitude Correlators) 3
1.3.2 Laplace Family (Time-Varying-Magnitude Correlators) 5
1.4 The DTFT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.5 Inverse DTFT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.6 A Few Properties of the DTFT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.6.1 Linearity 13
1.6.2 Conjugate Symmetry for real x[n] 13
1.6.3 Periodicity 13
1.6.4 Shift of Frequency 13
1.6.5 Convolution 14
1.6.6 Even and Odd Components 16
1.6.7 Multiplication By A Ramp 16
1.7 Frequency Response of an LTI System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.7.1 From Impulse Response 17
1.7.2 From Difference Equation 18
x CONTENTS
1.8 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.9 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

2 The z-Transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

2.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.2 Software for Use with this Book . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.3 Definition & Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.3.1 The z-Transform 26
2.3.2 The Inverse z-Transform 26
2.3.3 Convergence Criteria 26
2.3.4 Summary of ROC Facts 32
2.3.5 Trivial Poles and Zeros 33
2.3.6 Basic Properties of the z-Transform 33
2.3.7 Common z-Transforms 35
2.3.8 Transfer Functions, Poles, and Zeros 36
2.3.9 Pole Location and Stability 42
2.4 Conversion from z-Domain to Time Domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
2.4.1 Difference Equation 45
2.4.2 Table Lookup 46
2.4.3 Partial Fraction Expansion 46
2.4.4 Contour Integration in the Complex Plane 48
2.5 Transient and Steady-State Responses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
2.6 Frequency Response From z-Transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .55
2.6.1 For Generalized Transfer Function 55
2.6.2 Relation to DTFT 55
2.6.3 Finite Impulse Response (FIR) 56
2.6.4 Infinite Impulse Response (IIR) Single Pole 63
2.6.5 Cascaded Single-Pole Filters 64
2.6.6 Off-Unit-Circle Zeros and Decaying Signals 69
2.7 Transfer Function & Filter Topology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
2.7.1 Direct Form 70
2.7.2 Direct Form Transposed 71
2.7.3 Cascade Form 71
2.7.4 Parallel Form 76
2.7.5 Lattice Form 79
 CONTENTS xi
2.8 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
2.9 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

3 The DFT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93

3.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
3.2 Software for Use with this Book . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
3.3 Discrete Fourier Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
3.4 Sampling in the z-Domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
3.5 From DFS to DFT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
3.6 DFT-IDFT Pair . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
3.6.1 Definition-Forward Transform (Time to Frequency) 101
3.6.2 Definition-Inverse Transform (Frequency to Time) 101
3.6.3 Magnitude and Phase 101
3.6.4 N, Scaling Constant, and DFT Variants 101
3.7 MathScript Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .102
3.8 A Few DFT Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
3.9 General Considerations and Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
3.9.1 Bin Values 106
3.9.2 Periodicity in n and k 107
3.9.3 Frequency Multiplication in Time Domain 109
3.10 Computation of DFT Via Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .109
3.11 DFT of Common Signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
3.12 Frequency Resolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
3.13 Bin width and Sample Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
3.14 The FFT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
3.14.1 N-pt DFT from Two N/2-pt DFTs 121
3.14.2 Decimation-in-Time 123
3.14.3 Reassembly Via Butterfly 124
3.14.4 Algorithm Execution Time 126
3.14.5 Other Algorithms 127
3.15 The Goertzel Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
3.15.1 Via Single-Pole 128
3.15.2 Using Complex Conjugate Poles 130
3.15.3 Magnitude Only Output 130
3.16 Linear, Periodic, and Circular Convolution and the DFT . . . . . . . . . . . . . . . . . . . . . . . . .131
xii CONTENTS
3.16.1 Cyclic/Periodic Convolution 131
3.16.2 Circular Convolution 132
3.16.3 DFT Convolution Theorem 133
3.16.4 Linear Convolution Using the DFT 134
3.16.5 Summary of Convolution Facts 136
3.16.6 The Overlap-Add Method 136
3.17 DFT Leakage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
3.17.1 On-Bin/Off-Bin: DFT Leakage 140
3.17.2 Avoiding DFT Leakage-Windowing 141
3.17.3 Inherent Windowing by a Rectangular Window 142
3.17.4 A Few Common Window Types 144
3.17.5 DFT Leakage V. Window Type 146
3.17.6 Additional Window Use 150
3.18 DTFT Via Padded DFT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
3.19 The Inverse DFT (IDFT) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
3.20 Computation of IDFT Via Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
3.21 IDFT Via DFT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
3.22 IDFT Phase Descrambling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
3.22.1 Phase Zeroing 164
3.22.2 Phase Shifting 165
3.22.3 Equalization Using the DFT 165
3.23 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
3.24 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168

A Software for Use with this Book . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189

A.1 File Types and Naming Conventions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189

A.2 Downloading the Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
A.3 Using the Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
A.4 Single-line function calls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
A.5 Multi-line m-code examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .191
A.6 How to Successfully Copy-and-Paste M-code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .192
A.7 Learning To Use M-Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
A.8 What You Need with MATLAB and LabVIEW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193

B Vector/Matrix Operations in M-Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195

 CONTENTS xiii
B.1 Row and Column Vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
B.2 Vector Products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
B.2.1 Inner Product 195
B.2.2 Outer Product 195
B.2.3 Product of Corresponding Values 196
B.3 Matrix Multiplied by a Vector or Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
B.4 Matrix Inverse and Pseudo-Inverse . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .196

Biography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
 Preface to Volume II
0.1 INTRODUCTION
The present volume is Volume II of the series DSP for MATLAB ™ and LabVIEW ™ . The entire series
consists of four volumes which collectively form a work of twelve chapters that cover basic digital signal
processing in a practical and accessible manner, but which nonetheless include essential foundation
mathematics. The text is well-illustrated with examples involving practical computation using m-code
or MathScript (as m-code is usually referred to in LabVIEW-based literature), and LabVIEW VIs.
There is also an ample supply of exercises, which consist of a mixture of paper-and-pencil exercises
for simple computations, and script-writing projects having various levels of difficulty, from simple,
requiring perhaps ten minutes to accomplish, to challenging, requiring several hours to accomplish. As
the series title implies, the scripts given in the text and supplied in code form (available via the internet at
http://www.morganclaypool.com/page/isen) are suitable for use with both MATLAB (a product of The
Mathworks, Inc.), and LabVIEW (a product of National Instruments, Inc.). Appendix A in each volume
of the series describes the naming convention for the software written for the book as well as basics for
using the software with MATLAB and LabVIEW.

0.2 THE FOUR VOLUMES OF THE SERIES

The present volume, Volume II of the series, is devoted to discrete frequency transforms. It begins with
an overview of a number of well-known continuous domain and discrete domain transforms, and covers
the DTFT (Discrete Time Fourier Transform), the DFT (Discrete Fourier Transform), Fast Fourier
Transform (FFT), and the z-Transform in detail. Filter realizations (or topologies) are also covered,
including Direct, Cascade, Parallel, and Lattice forms.
Volume I of the series, Fundamentals of Discrete Signal Processing, consists of four chapters. The
first chapter gives a brief overview of the field of digital signal processing. This is followed by a chapter
detailing many useful signals and concepts, including convolution, recursion, difference equations, etc.
The third chapter covers conversion from the continuous to discrete domain and back (i.e., analog-to-
digital and digital-to-analog conversion), aliasing, the Nyquist rate, normalized frequency, conversion
from one sample rate to another, waveform generation at various sample rates from stored wave data, and
Mu-law compression. The fourth and final chapter of Volume I introduces the reader to many important
principles of signal processing, including correlation, the correlation sequence, the Real DFT, correlation
by convolution, matched filtering, simple FIR filters, and simple IIR filters.
Volume III of the series, Digital Filter Design, covers FIR and IIR design, including general
principles of FIR design, the effects of windowing and filter length, characteristics of four types of linear-
phase FIR, Comb and MA filters, Windowed Ideal Lowpass filter design, Frequency Sampling design
with optimized transition band coefficients, Equiripple FIR design, and Classical IIR design.
Volume IV of the series, LMS Adaptive Filtering, begins by explaining cost functions and perfor-
mance surfaces, followed by the use of gradient search techniques using coefficient perturbation, finally
reaching the elegant and computationally efficient Least Mean Square (LMS) coefficient update algo-
rithm. The issues of stability, convergence speed, and narrow-bandwidth signals are covered in a practical
xvi PREFACE
manner, with many illustrative scripts. In the second chapter of the volume, use of LMS adaptive fil-
tering in various filtering applications and topologies is explored, including Active Noise Cancellation
(ANC),system or plant modeling, periodic component elimination, Adaptive Line Enhancement (ADE),
interference cancellation, echo cancellation, and equalization/deconvolution.

0.3 ORIGIN AND EVOLUTION OF THE SERIES

The manuscript from which the present series of four books has been made began with an idea to provide
a basic course for intellectual property specialists and engineers that would provide more explanation and
illustration of the subject matter than that found in conventional academic books. The idea to provide an
accessible basic course in digital signal processing began in the mid-to-late 1990’s when I was introduced
to MATLAB by Dan Hunter, whose graduate school days occurred after the advent of both MATLAB
and LabVIEW (mine did not). About the time I was seriously exploring the use of MATLAB to update
my own knowledge of signal processing, Dr. Jeffrey Gluck began giving an in-house course at the agency
on the topics of convolutional coding, trellis coding, etc., thus inspiring me to do likewise in the basics of
DSP, a topic more in-tune to the needs of the unit I was supervising at the time. Two short courses were
taught at the agency in 1999 and 2000 by myself and several others, including Dr. Hal Zintel, David
Knepper, and Dr.Pinchus Laufer. In these courses we stressed audio and speech topics in addition to
basic signal processing concepts. Thanks to The Mathworks, Inc., we were able to teach the in-house
course with MATLAB on individual computers, and thanks to Jim Dwyer at the agency, we were able
to acquire several server-based concurrent-usage MATLAB licenses, permitting anyone at the agency to
have access to MATLAB. Some time after this, I decided to develop a complete course in book form, the
previous courses having consisted of an ad hoc pastiche of topics presented in summary form on slides,
augmented with visual presentations generated by custom-written scripts for MATLAB. An early draft
of the book was kindly reviewed by Motorola Patent Attorney Sylvia Y. Chen, which encouraged me to
contact Tom Robbins at Prentice-Hall concerning possible publication. By 2005, Tom was involved in
starting a publishing operation at National Instruments, Inc., and introduced me to LabVIEW with the
idea of possibly crafting a book on DSP to be compatible with LabVIEW. After review of a manuscript
draft by a panel of three in early 2006, it was suggested that all essential foundation mathematics be
included so the book would have both academic and professional appeal. Fortunately, I had long since
retired from the agency and was able to devote the considerable amount of time needed for such a project.
The result is a book suitable for use in both academic and professional settings, as it includes essential
mathematical formulas and concepts as well as simple or “first principle” explanations that help give the
reader a gentler entry into the more conventional mathematical treatment.
This double-pronged approach to the subject matter has, of course, resulted in a book of consider-
able length. Accordingly, it has been broken into four modules or volumes (described above) that together
form a comprehensive course, but which may be used individually by readers who are not in need of a
complete course.
Many thanks go not only to all those mentioned above, but to Joel Claypool of Morgan & Claypool,
Dr. C.L. Tondo and his troops, and, no doubt, many others behind the scenes whose names I have never
heard, for making possible the publication of this series of books.

Forester W. Isen
December 2008
 1

CHAPTER 1

The Discrete Time Fourier

Transform
1.1 OVERVIEW
1.1.1 IN THE PREVIOUS VOLUME
The previous volume of the series, Volume I, covered DSP fundamentals such as basic signals and LTI
systems, difference equations, sampling, the Nyquist rate, normalized frequency, correlation, convolution,
the real DFT, matched filtering, and basic IIR and FIR filters.

1.1.2 IN THIS VOLUME

In this volume, Volume II of the series, we take up discrete frequency transforms in detail, including an
overview of many transforms, both continuous-domain and discrete-domain, followed in sequence by
detailed discussions of a number of discrete transforms, knowledge of which is generally deemed essential
in the signal processing field.

1.1.3 IN THIS CHAPTER

We are now prepared in this chapter to begin a detailed exploration of discrete frequency transforms. A
number of such transforms exist, and we’ll begin by summarizing all of the basic facts and comparing each
to the others to better emphasize the important characteristics of each distinct transform. We include a
brief mention of continuous signal domain transforms for background and perspective, but concentrate
most of our effort on discrete signal transforms. All of the transforms we’ll investigate, both the continuous
and the discrete domain types, work on the same general concept–summing (or integrating in the case of
continuous time signals) the product of the signal and orthogonal-pair correlating waveforms of different
frequencies.
The transforms covered in detail in this book section are the Discrete Time Fourier Transform
(DTFT), which is covered in this chapter, the z-Transform (z-T), covered in the following chapter, and
the Discrete Fourier Transform (DFT), covered in the third and final chapter of this volume.
By the end of this chapter, the reader will have learned how to evaluate the frequency response
of an LTI system using the DTFT. This sets the stage for the following chapter, which discusses the
more generalized z-transform, which is in widespread use in industry and academia as the standard
method to describe the transfer function of an LTI system. This will be followed in the succeeding
chapter by a detailed look at the workhorse of practical frequency domain work, the DFT and a fast
implementation, the decimation-in-time FFT, as well as time domain convolution using the frequency
domain, the Goertzel algorithm, computing the DTFT using the DFT, etc.

1.2 SOFTWARE FOR USE WITH THIS BOOK

The software files needed for use with this book (consisting of m-code (.m) files, VI files (.vi), and related
support files) are available for download from the following website:
2 CHAPTER 1. THE DISCRETE TIME FOURIER TRANSFORM

http://www.morganclaypool.com/page/isen
The entire software package should be stored in a single folder on the user’s computer, and the full
file name of the folder must be placed on the MATLAB or LabVIEW search path in accordance with the
instructions provided by the respective software vendor (in case you have encountered this notice before,
which is repeated for convenience in each chapter of the book, the software download only needs to be
done once, as files for the entire series of four volumes are all contained in the one downloadable folder).
See Appendix A for more information.

1.3 INTRODUCTION TO TRANSFORM FAMILIES

The chief differences among the transforms mentioned below involve whether they 1) operate on con-
tinuous or discrete time signals, 2) provide continuous or discrete frequency output, and 3) use constant
unity-amplitude correlators (in the case of the Fourier family of transforms), or dynamic (decaying,
steady-state, or growing correlators) in the case of the Laplace and z- transforms.
The following table summarizes the main characteristics of a number of well-known transforms
with respect to the following categories: Input Signal Domain (continuous C or discrete D signal), Output
(Frequency) Domain (produces continuous C or discrete D frequency output), and Correlator Magnitude
(constant, unity magnitude for Fourier-based transforms, or variable magnitudes (decaying, growing, or
constant unity) for Laplace Transform and z-Transform, accordingly as e−σ t or |z|n , respectively.

Transform Input Output Correlator Mag.

Laplace Transform C C e−σ t

Fourier Transform C C 1
Fourier Series C D 1
Discrete Time Fourier Transform D C 1
Discrete Fourier Series D D 1
Discrete Fourier Transform D D 1
z-transform D C |z|n
For purposes of discrete signal processing, what is needed is a numerically computable repre-
sentation (transform) of the input sequence; that is to say, a representation which is itself a finite but
complete representation that can be used to reenter the time domain, i.e., reconstruct the original signal.
For transforms that are not computable in this sense, samples of the transform can be computed. Of all
the transforms discussed in the following section of the chapter, only the Discrete Fourier Series (DFS)
and Discrete Fourier Transform (DFT) are computable transforms in the sense mentioned above.
The use of dynamic correlators results in a transform that is an algebraic expression that implicitly
or explicitly contains information on the system poles and zeros. The system response to signals other
than steady-state, unity amplitude signals can readily be determined, although such transforms can also
be evaluated to produce the same result provided by the Fourier Transform (in the case of the Laplace
Transform) or the DTFT, DFS, and DFT (in the case of the z-Transform). Thus, the Laplace Trans-
form and z-Transform are more generalized transforms having great utility for system representation
and computation of response to many types of signals from the continuous and discrete time domains,
respectively.
While this book is concerned chiefly with discrete signal processing, we give here a brief discussion
of certain continuous time transforms (Laplace, Fourier, Fourier Series) to serve as background or points
 1.3. INTRODUCTION TO TRANSFORM FAMILIES 3
of reference for the discrete transforms that will be discussed in more detail below and in chapters to
follow.

1.3.1 FOURIER FAMILY (CONSTANT UNITY-MAGNITUDE CORRELATORS)

Fourier Transform  ∞
F (ω) = x(t)e−j ωt dt
−∞

The Fourier transform operates on continuous time, aperiodic signals and evaluates the frequency
response in the continuous frequency domain. The correlators are complex exponentials having constant
unity amplitude. Both t (time) and ω (frequency) run from negative infinity to positive infinity. The
Fourier Transform is a reversible transform; the inverse transform is
 ∞
x(t) = F (ω)ej ωt dω
−∞

Fourier Series
Many signals of interest are periodic, that is, they are composed of a harmonic series of cosines and sines.
For a periodic, continuous time signal of infinite extent in time, a set of coefficients can be obtained based
on a single period (between times to and to + T ) of the signal x(t):
 to +T
1
ck = x(t)e−j kω0 t dt
T to

where T is the reciprocal of the fundamental frequency F0 and k = 0, ±1, ±2, ...
For real x(t), ck and c−k are complex conjugates. If we say

ck = |ck | ej θk

then the original sequence can be reconstructed according to the formula


x(t) = c0 + 2 |ck | cos(2π kF0 t + θk )

An equivalent expression is

∞

x(t) = a0 + ak cos(2π kF0 t) − bk sin(2π kF0 t)
k=1

where a0 = c0 , ak = 2 |ck | cos θk , and bk = 2 |ck | sin θk .

Discrete Time Fourier Transform (DTFT)

The DTFT is defined for the discrete time input signal x[n] as

∞

DT F T (x[n]) = X(e ) = jω
x[n]e−j ωn (1.1)
n=−∞
4 CHAPTER 1. THE DISCRETE TIME FOURIER TRANSFORM
where ω (radian frequency) is a continuous function and runs from −π to π , and x[n] is absolutely
summable, i.e.,
∞
|x[n]| < ∞
n=−∞

The inverse DTFT (IDTFT) is defined as

 π
1
x[n] = X(ej ω )ej ωn dω (1.2)
2π −π

DTFT theory will be discussed in detail below, while computation of the DTFT using the DFT
will be discussed in the chapter on the DFT.

Discrete Fourier Series (DFS)

A periodic sequence x[n] (-∞ < n < ∞) may be decomposed into component sequences that comprise
a harmonic series of complex exponentials. Since a sampled sequence is bandlimited, it follows that the
frequency of the highest harmonic is limited to the Nyquist rate. The normalized frequencies of the
harmonics are 2πk/N , with k = 0:1:N -1. Since the transform involves a finite number of frequencies to
be evaluated, the DFS is a computable transform.
 are
The DFS coefficients X[k]


N−1
 =
DF S(x[n]) = X[k] x [n]e−j 2πkn/N
 (1.3)
n=0

where k is an integer and 

x [n] is one period of the periodic sequence x[n]. Typical ranges for k are:
0:1:N -1, or -N/2+1:1:N/2 for even length sequences, or -(N -1)/2:1:(N -1)/2 for odd length sequences.
Once having the DFS coefficients, the original sequence x[n] can be reconstructed using the
following formula:

1 
N−1
x[n] = X[k]ej 2πkn/N (1.4)
N
k=0

The DFS, a computable transform, forms an important theoretical basis for the Discrete Fourier
Transform and will be discussed in more detail in the chapter on the DFT.

Discrete Fourier Transform (DFT)

1 
N−1
DF T [k] = x[n]e−j 2πkn/N
N
n=0

The DFT (a computable transform) operates on discrete, or sampled signals, and evaluates fre-
quency response at a number (equal to about half the sequence length) of unique frequencies. The
correlators are complex exponentials having constant, unity amplitude, and the frequencies k range from
0 to N − 1 or, for even-length sequences, −N/2 + 1 to N/2 and −N/2 to −N/2 for odd-length DFTs.
The DFT (including its efficient implementation, the FFT) will be discussed extensively in the
chapter that follows the chapter on the z-transform.
 1.3. INTRODUCTION TO TRANSFORM FAMILIES 5
1.3.2 LAPLACE FAMILY (TIME-VARYING-MAGNITUDE CORRELATORS)
Laplace Transform (LT)
The LT is defined as
 ∞  ∞
£(s) = x(t)e−st dt = x(t)e−σ t e−j ωt dt
−∞ −∞

The LT is the standard frequency transform for use with continuous time domain signals and
systems. The parameter s represents the complex number σ + j ω, with σ , a real number, being a damping
coefficient, and j ω, an imaginary number, representing frequency. Both σ and ω run from negative infinity
to positive infinity. The correlators generated by e−st are complex exponentials having amplitudes that
decay, grow, or retain unity-amplitude over time, depending on the value of σ . By varying both σ and ω,
the poles and zeros of the signal or system can be identified. Results are graphed in the s-Domain (the
complex plane), using rectangular coordinates with σ along the horizontal axis, and j ω along the vertical
axis. The magnitude of the transform can be plotted along a third dimension in a 3-D plot if desired, but,
more commonly, a 2-D plot is employed showing only the locations of poles and zeros.
The LT is a reversible transform, and can be used to solve differential equations, such as those
representing circuits having inductance and capacitance, in the frequency domain. The time domain
solution is then obtained by using the Inverse LT. The LT is used extensively for circuit analysis and
representation in the continuous domain. We’ll see later in the book that certain well known or classical
IIRs (Butterworth, Chebyshev, etc) have been extensively developed in the continuous domain using
LTs, and that the Laplace filter parameters can be converted to the digital domain to create an equivalent
digital IIR.
Note that the FT results when σ = 0 in the Laplace transform. That is to say, when the damping
coefficient is zero, the Laplace correlators are constant, unity-amplitude complex exponentials just as
those of the FT. Information plotted along the Imaginary axis in the s- or Laplace domain is equivalent,
then, to the FT.

z-Transform (z-T)
The z-transform (z-T) is a discrete time form of the Laplace Transform. For those readers not familiar
with the LT, study of the z-T can prove helpful since many Laplace properties and transforms of common
signals are analogous to those associated with the z-T. The z-T is defined as
∞

X(z) = x[n]z−n
n=−∞

The z-T converts a number sequence into an algebraic expression in z, and, in the reverse or inverse
z-T, converts an algebraic expression in z into a sequence of numbers.The z-T is essentially a discrete-time
version of the Laplace transform. The correlators are discrete-time complex exponentials with amplitudes
that grow, shrink, or stay the same according to the value of z (a complex number) at which the transform
is evaluated. Values of z having magnitudes < 1 result in a correlation of the signal with a decaying
(discrete) complex exponential, evaluation with z having a magnitude equal to 1 results in a Fourier-like
response, and evaluation with z having a magnitude greater than 1 results in a correlation of the signal with
a growing discrete complex exponential. Results can be plotted in the z-Domain, which is the complex
plane using polar coordinates of r and θ, where θ corresponds to normalized radian frequency and r to a
damping factor. The unit circle in the z-domain corresponds to the imaginary axis in the s-Domain; the
left hand s-plane marks a region of stable pole values in the s-plane that corresponds to the area inside
6 CHAPTER 1. THE DISCRETE TIME FOURIER TRANSFORM
the unit circle in the z-plane. As in the s-plane, the z-T magnitude may be plotted using a 3-D plot, or,
using a 2-D plot, only the location of the poles and zeros may be plotted.
Among the discrete-signal transforms, the z-T is more general than the DTFT, since in the z-T,
the test exponentials may also have decay factors (negative or positive). All of the members of the Fourier
transform family use exponentials of constant unity amplitude to perform their correlations with the
signal of unknown frequency content, and thus cannot give pole and zero locations as can the Laplace
and z- transforms.
The z-transform will be discussed extensively in the next chapter.
Additional Transforms
There are also several other transforms that are derived from the DFT; namely, the Discrete Cosine
Transform (DCT), and the Discrete Sine Transform (DST). These transforms use multiples of half-
cycles of either the cosine or sine, respectively, as the correlators, rather than multiples of full cycles as in
the DFT. The bin values are real only. A form of the DCT called the MDCT (Modified DCT) is used
in certain audio compression algorithms such as MP3.
Reference [1] gives a thorough and very accessible development of the DFT, and also briefly
discusses the Laplace transform, the z-transform, the DCT, and the DST.

1.4 THE DTFT

The DTFT provides a continuous frequency spectrum for a sampled signal, as opposed to a discrete
frequency spectrum (in which only a finite number of frequency correlators are used, as is true of the
DFT and DFS).
In Eq. (1.1), x[n] is a sampled signal which may be of either finite or infinite extent, and ω, a
continuous function of frequency which may assume values from −π to π .

Example 1.1. Derive an algebraic expression for the DTFT of the function 0.9n u[n], and then evaluate
it numerically at frequencies between 0 and π radians, at intervals of 0.01 radian.

We must evaluate the expression

∞

DT F T (x[n]) = x[n]e−j ωn
n=−∞

We note that x[n] = 0 for n < 0, that x[n] itself is absolutely summable since it is a decreasing
geometric series. The n-th term of the summation is

0.9n e−j ωn
and we note that each successive term is arrived at by multiplying the previous term by the number
0.9e−j ω and thus the net sequence forms a geometric series the sum of which is
1
(1.5)
(1 − 0.9e−j ω )
We can evaluate this expression at a finite number of values of ω and plot the result. The following
code directly evaluates expression (1.5) at frequencies between 0 and π radians, as specified by the vector
w:
 1.4. THE DTFT 7
w = 0:0.01:pi; DTFT = 1./(1-0.9*exp(-j*(w)));
figure(8); plot(w/(pi),abs(DTFT));
xlabel(’Normalized Frequency’); ylabel(’Magnitude’)

Example 1.2. Evaluate and plot the magnitude of the DTFT of the following sequence: [1, 0, 1].


2 
F (ω) = x[n]e−j ωn = [1, 0, 1][e−j ω0 , e−j ω1 , e−j ω2 ] = 1 + e−j ω2 (1.6)
n=0

From our earlier work, we recognize the impulse response [1,0,1] as that of a simple bandstop
filter. We can show that this is so by evaluating Eq. (1.6) at a large (but necessarily finite) number of
values of ω with the following code, the results of which are shown in Fig. 1.1.
w = 0:0.01:pi; DTFT = 1+exp(-j*2*w);
figure(8); plot(w/(pi),abs(DTFT));
xlabel(’Normalized Frequency’); ylabel(’Magnitude’)

1.8

1.6

1.4

1.2
Magnitude

0.8

0.6

0.4

0.2

0
0 0.2 0.4 0.6 0.8 1
Normalized Frequency

Figure 1.1: Magnitude of the DTFT of the simple notch filter [1, 0, 1].
8 CHAPTER 1. THE DISCRETE TIME FOURIER TRANSFORM
Example 1.3. Write a script that will evaluate and plot the magnitude and phase of the DTFT for any
sequence; test it for the sequences [1, 0, 1], [1, 0,-1], and [1, 0, 0, 1].

Such a script should allow one to specify how many samples M of the DTFT to compute over the
interval 0 to Rπ, with R = 1 being suitable for real x[n] and R = 2 being suitable for complex x[n]. Values
of R greater than 2 allow demonstration of periodicity of the DTFT. The code creates an n-by-k matrix
dMat of complex correlators, where each column is a complex correlator of length n and frequency k.
The DTFT is obtained by multiplying the signal vector x on the right by dMat. Each element in the
resulting row vector of frequency responses is obtained as the inner or dot product of the signal vector x
with a column of dMat.
function LV_DTFT_Basic(x,M,R)
% LV_DTFT_Basic([1,0,1],300,1)
N = length(x); W = exp(-j*R*pi/M); k = 0:1:M-1;
n = 0:1:N-1; dMat = W.ˆ(n’*k); d = x*dMat; figure(9)
subplot(2,2,1); plot(R*[0:1:M-1]/M,abs(d));
grid; xlabel(’Norm Freq’); ylabel(’Mag’)
subplot(2,2,2); plot(R*[0:1:M-1]/M,angle(d))
grid; xlabel(’Norm Freq’); ylabel(’Radians’)
subplot(2,2,3); plot(R*[0:1:M-1]/M,real(d));
grid; xlabel(’Norm Freq’); ylabel(’Real’)
subplot(2,2,4); plot(R*[0:1:M-1]/M,imag(d))
grid; xlabel(’Norm Freq’); ylabel(’Imag’)
The result from making the call

LV_DTFT_Basic([1,0,1],300,1)
is shown in Fig. 1.2.
A more versatile version of the above code is the script (see exercises below)

LV xDT F T (x, n, M, R, F reqOpt, F igN o)

which, from sequence x having time indices n, computes M frequency samples over the interval Rπ ,
which can be computed symmetrically or asymmetrically with respect to frequency zero (FreqOpt = 1 for
symmetrical, 2 for asymmetrical). The radian frequencies of evaluation would be, for the asymmetrical
option

Rπ([0 : 1 : M − 1])/M
and for the symmetrical option

Rπ([−(M − 1)/2 : 1 : (M − 1)/2])/M ( M odd )

Rπ([−M/2 + 1 : 1 : M/2])/M ( M even )

The desired figure number is supplied as FigNo, an option allowing you to create different figures
for comparison with each other. A typical call is
 1.4. THE DTFT 9

2 2

1
1.5

Radians
0
Mag

1
−1
0.5
−2

0 −3
0 0.5 1 0 0.5 1
(a) Norm Freq (b) Norm Freq
2 2

1.5
1
1
Imag
Real

0
0.5
−1
0

−0.5 −2
0 0.5 1 0 0.5 1
(c) Norm Freq (d) Norm Freq

Figure 1.2: (a) Magnitude of DTFT of the sequence [1 0 1]; (b) Phase response of DTFT; (c) Real
component of DTFT ; (d) Imaginary component of DTFT.

LVxDTFT([1,0,1],[0:1:2],300,2,1,88)
which results in Fig. 1.3.
A second script (for a complete description of input arguments, see exercises below)

LV xDT F T _MS(x, SampOff set, F reqOff setExp, M, R, T imeOpt, F reqOpt)

allows you to enter one sequence, and the second sequence is created as a modification of the first, delayed
by SampOffset samples and offset in frequency by the complex exponential FreqOffsetExp. Input arguments
M and R are as described for the script LVxDTFT ; F reqOpt determines whether the DTFT is computed
symmetrically or asymmetrically about frequency zero, as described above for the script LVxDTFT. For
the asymmetrical time option (determined by the input argument TimeOpt), the sequence time indices
of the first sequence are given as

n = 0 : 1 : N − 1;
10 CHAPTER 1. THE DISCRETE TIME FOURIER TRANSFORM

4
Mag

2
0
−1 −0.5 0 0.5 1
Normalized Frequency (Multiples of π)
5
Radians

0
−5
−1 −0.5 0 0.5 1
Normalized Frequency (Multiples of π)
4
Real

2
0
−1 −0.5 0 0.5 1
Normalized Frequency (Multiples of π)
4
Imag

2
0
−1 −0.5 0 0.5 1
Normalized Frequency (Multiples of π)

Figure 1.3: (a) Magnitude of DTFT of the sequence [1,0,1]; (b) Phase of DTFT of same; (c) Real part
of DTFT of same; (d) Imaginary part of DTFT of same.

where N is the length of x. For the symmetrical time index option, the time indices are given as

n = −(N − 1)/2 : 1 : (N − 1)/2 (n odd)

n = −N/2 + 1 : 1 : N/2 (n even)

This script is useful for demonstrating the effect on the DTFT of time and frequency shifts to a
first test sequence. A typical call, which results in Fig. 1.4, is
nN = (0:1:100)/100;
LVxDTFT_MS([cos(2*pi*25*nN)],0,...
exp(j*2*pi*12.5*nN),200,2,2,1)
We will make use of these scripts shortly while studying the various properties of the DTFT.
 1.5. INVERSE DTFT 11

100 100
Mag

Mag
50 50
0 0
−1 −0.5 0 0.5 1 −1 −0.5 0 0.5 1
(a) Norm Freq, x1[n] (e) Norm Freq, x2[n]
2 2
Radians

Radians
0 0
−2 −2
−1 −0.5 0 0.5 1 −1 −0.5 0 0.5 1
(b) Norm Freq, x1[n] (f) Norm Freq, x2[n]
100 100
Real

Real
0 0
−100 −100
−1 −0.5 0 0.5 1 −1 −0.5 0 0.5 1
(c) Norm Freq, x1[n] (g) Norm Freq, x2[n]
100 100
Imag

Imag

0 0
−100 −100
−1 −0.5 0 0.5 1 −1 −0.5 0 0.5 1
(d) Norm Freq, x [n] (h) Norm Freq, x [n]
1 2

Figure 1.4: DTFT of first sequence, with its magnitude and phase and real and imaginary parts being
shown, respectively, in plots (a)-(d); DTFT of second sequence, which is the first sequence offset in
frequency by π/4 radian, with its magnitude, phase, and real and imaginary parts being shown, respectively,
in plots (e)-(h). All frequencies are normalized, i.e., in units of π radians.

1.5 INVERSE DTFT

The Inverse DTFT, i.e., the original time domain sequence x[n] from which a given DTFT was produced,
can be reconstructed by evaluating the following integral:
 π
1
x[n] = X(ej ω )ej ωn dω (1.7)
2π −π
We will illustrate this with several examples, one analytic and the other numerical.

Example 1.4. Using Eq. (1.7), compute x[0], x[1], and x[2] from the DTFT obtained in Eq. (1.6).

We get
12 CHAPTER 1. THE DISCRETE TIME FOURIER TRANSFORM

 π
1
x[0] = (1 + e−j ω2 )ej ω0 dω
2π −π
and
 π
1
x[1] = (1 + e−j ω2 )ej ω1 dω
2π −π
and
 π
1
x[2] = (1 + e−j ω2 )ej ω2 dω
2π −π
The formula for x[0] reduces to
 π  π  π
1 1
(1 + e−j ω2 )dω = ( dω + e−j ω2 dω)
2π −π 2π −π −π
which is
 π
1
(ω |π−π + e−j ω2 dω) = 1 + 0 = 1
2π −π
where we note that
 π 
2π if n=0
e±j ωn dω =
−π 0 if n = ±1, ±2...
For x[1] we get
 π  π
1 1
x[1] = (1 + e−j ω2 )ej ω1 dω = (ej ω1 + e−j ω1 )dω = 0
2π −π 2π −π
The formula for x[2] is the same as that for x[0] with the exception of the sign of the complex
exponential, which does not affect the outcome. The reconstructed sequence is therefore [1,0,1], as
expected.

Example 1.5. Using numerical integration, approximate the IDTFT that was determined analytically
in the previous example.

Let’s reformulate the code to obtain the DTFT from -pi to +pi, and to use a much finer sample
spacing (this will improve the approximation to the true, continuous spectrum DTFT), and then perform
the IDTFT one sample at a time in accordance with Eq. (1.2):
N=10ˆ3; dw = 2*pi/N; w = -pi:dw:pi*(1-2/N);
DTFT = 1+exp(-j*2*w);
x0 = (1/(2*pi))*sum(DTFT.*exp(j*w*0)*dw)
x1 = (1/(2*pi))*sum(DTFT.*exp(j*w*1)*dw)
x2 = (1/(2*pi))*sum(DTFT.*exp(j*w*2)*dw)
Running the preceding code yields the following answer, which rounds to the original sequence,
[1,0,1]:
 1.6. A FEW PROPERTIES OF THE DTFT 13
x0 = 1.0000 - 0.0000i
x1 = -4.9127e-017
x2 = 1.0000 + 0.0000i

1.6 A FEW PROPERTIES OF THE DTFT

1.6.1 LINEARITY
The DTFT of a linear combination of two sequences x1 [n] and x2 [n] is equal to the sum of the individual
responses, i.e.,
DT F T (ax1 [n] + bx2 [n]) = aDT F T (x1 [n]) + bDT F T (x2 [n])

1.6.2 CONJUGATE SYMMETRY FOR REAL x[n]

For real x[n], the real part of the DTFT shows even symmetry (X(ej ω ) = X(e−j ω )), and the imaginary
part shows odd symmetry (X(ej ω ) = −X(e−j ω )).

1.6.3 PERIODICITY
The DTFT of a sequence x[n] repeats itself every 2π :

X(ej w ) = X(ej (w+2πn) )

where n = 0, ±1, ±2...
To illustrate this principle, consider the following: for a given sequence length, the Nyquist limit is
half the sequence length, and this represents a frequency shift of π radians. To shift 2π radians therefore
is to shift by a frequency equal to the sequence length. The following m-code, the results of which are
illustrated in Fig. 1.5, verifies this property.
SR = 100; nN = (0:1:SR)/SR;
LVxDTFT_MS([cos(2*pi*25*nN)],0,exp(j*2*pi*SR*nN),200,2,2,1)
You can gain insight by noting that the code

nN = (0:1:100)/100; y = exp(j*2*pi*100*nN)
yields y = ones(1, 101), which clearly transforms the original sequence into itself, i.e., the new sequence
is the same as the old, and hence the DTFT is the same. In other words, the DTFT of a sequence repeats
itself for every frequency shift of 2π radians of the original sequence.

1.6.4 SHIFT OF FREQUENCY

If the signal x[n] is multiplied by a complex exponential of frequency F0 , the result is that the DTFT of
x[n] is shifted.

DT F T (x[n]ej ω0 n ) = X(ej (ω−ω0 ) )

To demonstrate this property, we can use the script LVxDTFT_MS. We pick the short sequence
[1,0,1] as x[n], and specify no sample offset, but a frequency offset of 2π/16 radians. We thus make the
call

LVxDTFT_MS([1,0,1],0,exp(j*2*pi*1*(0:1:2)/16),100,2,1,2)
14 CHAPTER 1. THE DISCRETE TIME FOURIER TRANSFORM

100 100
Mag

Mag
50 50
0 0
−1 −0.5 0 0.5 1 −1 −0.5 0 0.5 1
(a) Norm Freq, x [n] (e) Norm Freq, x [n]
1 2
2 2
Radians

Radians
0 0
−2 −2
−1 −0.5 0 0.5 1 −1 −0.5 0 0.5 1
(b) Norm Freq, x [n] (f) Norm Freq, x [n]
1 2
100 100
Real

Real
0 0
−100 −100
−1 −0.5 0 0.5 1 −1 −0.5 0 0.5 1
(c) Norm Freq, x1[n] (g) Norm Freq, x2[n]
100 100
Imag

Imag

0 0
−100 −100
−1 −0.5 0 0.5 1 −1 −0.5 0 0.5 1
(d) Norm Freq, x1[n] (h) Norm Freq, x2[n]

Figure 1.5: DTFT of first sequence, with its magnitude and phase and real and imaginary parts being
shown, respectively, in plots (a)-(d); DTFT of second sequence, which is the first sequence offset in
frequency by 2π radians, with its magnitude, phase, and real and imaginary parts being shown, respectively,
in plots (e)-(h). All frequencies are normalized, i.e., in units of π radians.

which results in Fig. 1.6. The reader should be able to verify by visual comparison of plots (a) and (e) that
the magnitude of frequency response has in fact been shifted by π/8 radian.

1.6.5 CONVOLUTION
The DTFT of the time domain convolution of two sequences is equal to the product of the DTFTs of
the two sequences.

DT F T (x1 [n] ∗ x1 [n]) = X1 (ej ω )X2 (ej ω )

Example 1.6. Consider the two sequences [1, 0, 1] and [1, 0, 0, −1]. Obtain the time domain convo-
lution by taking the inverse DTFT of the product of the DTFTs of each sequence, and confirm the result
using time domain convolution.
 1.6. A FEW PROPERTIES OF THE DTFT 15

2 2
Mag

Mag
1 1
0 0
0 0.5 1 1.5 2 0 0.5 1 1.5 2
(a) Norm Freq, x [n] (e) Norm Freq, x [n]
1 2
5 5
Radians

Radians
0 0
−5 −5
0 0.5 1 1.5 2 0 0.5 1 1.5 2
(b) Norm Freq, x1[n] (f) Norm Freq, x2[n]
2 2
Real

Real
0 0
−2 −2
0 0.5 1 1.5 2 0 0.5 1 1.5 2
(c) Norm Freq, x1[n] (g) Norm Freq, x2[n]
2 2
Imag

Imag

0 0
−2 −2
0 0.5 1 1.5 2 0 0.5 1 1.5 2
(d) Norm Freq, x1[n] (h) Norm Freq, x2[n]

Figure 1.6: DTFT of first sequence, with its magnitude and phase and real and imaginary parts being
shown, respectively, in plots (a)-(d); DTFT of second sequence, which is the first sequence offset in
frequency by π/8 radian, with its magnitude, phase, and real and imaginary parts being shown, respectively,
in plots (e)-(h). All frequencies are normalized, i.e., in units of π radians.

The DTFT of the first sequence is

X1 (ej ω ) = 1 + e−j ω2
and for the second sequence

X2 (ej ω ) = 1 − e−j ω3
The product is

1 + e−j ω2 − e−j ω3 − e−j ω5

or stated more completely
16 CHAPTER 1. THE DISCRETE TIME FOURIER TRANSFORM

1e−j ω0 + 0e−j ω1 + e−j ω2 − e−j ω3 + 0e−j ω4 − e−j ω5

which can be recognized by inspection as the DTFT of the time domain sequence

[1, 0, 1, −1, 0, −1]

To confirm, we make the following call:

y = conv([1,0,1],[1,0,0,-1])
which produces the same result.

1.6.6 EVEN AND ODD COMPONENTS

If a real sequence x[n] is decomposed into its even and odd components, and the DTFT taken of each,
it will be found that the DTFT of the even part is all real and equal to the real part of DT F T (x[n]),
while the DTFT of the odd part will be found to be imaginary only and equal to the imaginary part of
DT F T (x[n]). Using X(ej ω ), XE(ej ω ), XO(ej ω ) as the DTFT’s of x[n], xe[n], and xo[n], we would
have

Re(X(ej ω )) = Re(XE(ej ω )) = XE(ej ω )

and

Im(X(ej ω )) = Re(XO(ej ω )) = XO(ej ω )

Example 1.7. Write a script that will demonstrate the above even-odd properties.

We can perform the even-odd decomposition using the script LVEvenOddSymmZero presented
earlier, and then use the script LVxDTFT in three separate calls to open three separate windows to show
X(ej ω ), XE(ej ω ), and XO(ej ω ); the following script also reconstructs the original DTFT as the sum
of its even and odd components, subtracts this from the DTFT of the original signal, and obtains the
RMS error, which should prove to be essentially zero, within the limits of roundoff error.
x = [1:1:9]; [xe,xo,m] = LVEvenOddSymmZero(x,[0:1:8]);
d = LVxDTFT(x,[0:1:8],200,2,2,10);
de = LVxDTFT(xe,m,200,2,2,11);
do = LVxDTFT(xo,m,200,2,2,12);
RMS = sqrt((1/200)*sum((d - (de+do)).ˆ2))
The value of RMS reported after running the above code was [1.0190e-016, -1.6063e-015i], which
is essentially zero, within the limits of roundoff error.

1.6.7 MULTIPLICATION BY A RAMP

dX(ej ω )
DT F T (nx[n]) = j
dω
This property states that if the DTFT of x[n] is X(ej ω ), then the DTFT of nx[n] is the derivative
with respect to ω of X(ej ω ) multiplied by j .
 1.7. FREQUENCY RESPONSE OF AN LTI SYSTEM 17

1.7 FREQUENCY RESPONSE OF AN LTI SYSTEM

1.7.1 FROM IMPULSE RESPONSE
An interesting and useful result occurs by convolving a complex exponential ej ωk of radian frequency of
ω = ω0 with an LTI system having an impulse response represented by h[n]:
∞
 
y[k] = h[n]ej ω0 (k−n) = ( h[n]e−j ω0 n )ej ω0 k (1.8)
n=−∞

The rightmost expression in Eq. (1.8) is the input signal ej ω0 k scaled by the DTFT of h[n] evaluated at
ω0 . Since the DTFT evaluated at a single frequency is a complex number, it is sometimes convenient to
represent it as a magnitude and phase angle. Thus, for each frequency ω0 we would have the result

y[k] = H (ej ω0 )ej ω0 k = (|H (ej ω0 )|)( H (ej ω0 ))ej ω0 k (1.9)

An interpretation of this result is that a sinusoidal excitation of frequency ω0 to an LTI system
produces an output that is a sinusoid of the same frequency, scaled by the magnitude of the DTFT of
h[n] at ω0 , and phase shifted by the angle of the DTFT of h[n] evaluated at ω0 . This is consistent with
the principle of Sinusoidal Fidelity discussed in Volume I of this series.

Example 1.8. Consider the LTI system whose impulse response is [1, 0, 1]. Determine the magnitude
of the response to the complex exponential.

exp(j*2*pi*(0:1:31)*5/32)
We note that the normalized frequency is 5/16 = 0.3125 and obtain the magnitude of H at the
given frequency as 1.111 using the following code:
w = 5*pi/16; DTFT = 1+exp(-j*2*w);
magH = abs(DTFT)
To obtain the magnitude of response via time domain convolution, we make the following call:
tdMagResp = max(abs(conv([1,0,1],exp(j*2*pi*(0:1:31)*5/32))))
which produces the identical result, 1.111.
Equation (1.9) may be generalized and applied to real sinusoids. Thus, the steady state response
y[n] of an LTI system to a cosine (or sine) of magnitude A, frequency ω0 and phase angle φ, with the
DTFT magnitude and angle being M and θ is

y[n] = MA cos(ω0 n + φ + θ) (1.10)

Example 1.9. Verify Eq. (1.10) using both cosine and sine waves having normalized frequencies of 0.53
for the LTI system whose impulse response is [1, 0, 1].

We’ll use a discrete cosine of the stated frequency with φ = 0 and M = 1, and then compute the
output according to Eq. (1.10) and then by convolution with the impulse response. The plot will show
that the results are identical except for the first two and last two samples (recall that Eq. (1.10) represents a
18 CHAPTER 1. THE DISCRETE TIME FOURIER TRANSFORM
steady state response). In the code below, you can substitute the sine function (sin) for the cosine function
(cos) as well as change the frequency F and test signal length N .
F = 0.53; N = 128; w = F*pi; dtft = 1+exp(-j*2*w);
M = abs(dtft); theang = angle(dtft);
y1 = M*cos(2*pi*(0:1:N-1)*(N/2*F)/N + theang);
y2 = conv([1 0 1],cos(2*pi*(0:1:N-1)*(N/2*F)/N));
stem(y1,’bo’); hold on; stem(y2,’r*’)
Since LTI systems obey the law of superposition, Eq. (1.10) can be generalized for an input signal
comprising a sum or superposition of sinusoids:


K
y[n] = Mk Ak cos(ωk n + φk + θk ) (1.11)
k=1

1.7.2 FROM DIFFERENCE EQUATION

Thus, far we have been using the impulse response of an LTI system to determine the system’s frequency
response; in many cases, a difference equation may be what is immediately available. While it is certainly
possible to process a unit impulse using the difference equation and thus obtain the impulse response,
it is possible to obtain the frequency response directly from the difference equation. Assuming that the
difference equation is of the form


K 
M
y[n] + ak y[n − k] = bm x[n − m] (1.12)
k=1 m=0

and recalling from Eq. (1.9) that

y[n] = H (ej ω )ej ωn

and substituting in Eq. (1.12), we get


K 
M
H (ej ω )ej ωn + ak H (ej ω )ej ω(n−k) = bm ej ω(n−m)
k=1 m=0

which reduces to
M −j ωm
m=0 bm e
H (ej ω ) = K (1.13)
1 + k=1 ak e−j ωk

Example 1.10. Determine the frequency response of a certain LTI system that is defined by the following
difference equation, then compute and plot the magnitude and phase response for ω = 0 to 2π .

y[n] = x[n] + x[n − 2] + 0.9y[n − 1]

We first rewrite the equation to have all y terms on the left and all x terms on the right:
 1.8. REFERENCES 19

y[n] − 0.9y[n − 1] = x[n] + x[n − 2]

Then we get

1 + e−j ω2
H (ej ω ) =
1 − 0.9e−j ω
which can be numerically evaluated using the following code:
N=10ˆ3; dw=2*pi/N; w = 0:dw:2*pi-dw;
H = (1+exp(-j*2*w))./(1-0.9*exp(-j*w));
figure; subplot(2,1,1); plot(abs(H)); subplot(2,1,2); plot(angle(H))

1.8 REFERENCES
[1] William L. Briggs and Van Emden Henson, The DFT, An Owner’s Manual for the Discrete Fourier
Transform, SIAM, Philadelphia, 1995.
[2] Steven W. Smith, The Scientist and Engineer’s Guide to Digital Signal Processing, California Technical
Publishing, San Diego, 1997.
[3] Alan V. Oppenheim and Ronald W. Schaefer, Discrete-Time Signal Processing, Prentice-Hall, Engle-
wood Cliffs, New Jersey, 1989.
[4] John G. Proakis and Dimitris G. Manolakis, Digital Signal Processing, Principles, Algorithms, and
Applications, Third Edition, Prentice Hall, Upper Saddle River, New Jersey, 1996.
[5] Vinay K. Ingle and John G. Proakis, Digital Signal Processing Using MATLAB V.4, PWS Publishing
Company, Boston, 1997.

1.9 EXERCISES
1. Determine analytically the DTFT of the following sequences:
(a) 0.95n u[n] - 0.8n−1 u[n − 1]
(b) cos(2π (0:1:3)/4)
(c) (0.85)n−2 u[n − 2]
(d) ([0,0,(0.85)n−2 u[n − 2]])u[n]
(e) (-0.9)n u[n] + (0.7)n−3 u[n − 3]
(f ) n(u[n] - u[n − 3])
2. Compute the DTFTs of the following sequences and plot the magnitude and phase responses (use
one of the scripts developed or presented earlier in this chapter, such as LV_DTFT_Basic). Be sure to
increase the number of DTFT samples as the signal length increases. It’s a good idea to use at least 10
times as many DTFT samples as the sequence length being evaluated. Consider plotting the magnitude
on a logarithmic scale to see fine detail better. To avoid the problem of taking the logarithm of zero, add a
small number such as 10−10 to the absolute value of the DTFT, then use the function log10 and multiply
the result by 20.
(a) cos(2*pi*k*(0:1:N-1)/N) for N = 8, 32, 128 and k = 0, 1, and N/2.
20 CHAPTER 1. THE DISCRETE TIME FOURIER TRANSFORM
(b) [ones(1,10),zeros(1,20),ones(1,10),zeros(1,20)]
(c) b = fir1(21,0.5)
(d) [1,0,1]
(e) [1,0,1,0,1,0,1]
(f ) [1,0,1,0,1,0,1,0,1,0,1,0,1]
(g) [1,0,1,0,1,0,1,0,1,0,1,0,1].*hamming(13)’
(h) [1,0,1,0,1,0,1,0,1,0,1,0,1].*blackman(13)’
(i) [1,0,1,0,1,0,1,0,1,0,1,0,1].*kaiser(13,5)’
(j) [1,0,-1,0,1,0,-1,0,1,0,-1,0,1].*hamming(13)’
(k) [1,0,-1,0,1,0,-1,0,1,0,-1,0,1].*blackman(13)’
(l) [1,0,-1,0,1,0,-1,0,1,0,-1,0,1].*kaiser(13,5)’
(m) [real(j.ˆ(0:1:10))].*blackman(11)’
(n) [real(j.ˆ(0:1:20))].*blackman(21)’
(o) [real(j.ˆ(0:1:40))].*blackman(41)’
(p) [real(j.ˆ(0:1:80))].*blackman(81)’
3. Write a script that can receive b and a difference equation coefficients (according to Eq. (1.13)) in row
vector form (normalized so a0 = 1), compute, and display the following:
(a) The unit impulse response of the system defined by b and a.
(b) The unit step response of the system.
(c) The magnitude and phase of the DTFT.
(d) The response to a linear chirp of length 1024 samples and frequencies from 0 to 512 Hz (0 to
π radians in normalized frequency).
(e) The magnitude of response to a complex linear chirp of length 1024 samples and frequencies
from 0 to 512 Hz (0 to π radians in normalized frequency). Such a chirp can be generated by the following
code:
N = 1024; t = 0:1/N:1; y = chirp(t,0,1,N/2) + ...
j*chirp(t,0,1,N/2,’linear’,90)
(f ) The response to a signal of length 1024 samples containing a cosine of frequency 128.
(g) The response to a signal of length 1024 samples containing a cosine of frequency 256.
Use the script to evaluate the LTI systems defined by the following difference equations or b and
a coefficients. You should note that some of the systems are not stable. Compare the plot of DTFT
magnitude to the two chirp responses for each of the difference equations below. Where can the steady-
state magnitude of the responses specified in (b), (f ), and (g) above be found on the DTFT magnitude
plot? State whether each of the systems is stable or unstable.
(I) y[n] = x[n] + x[n-1] + 0.9y[n-1]
(II) y[n] = x[n] + 1.4y[n-1] - 0.81y[n-2]
(III) y[n] = x[n] + 1.4y[n-1] + 0.81y[n-2]
(IV) y[n] = x[n] - 2.45y[n-1] +2.37y[n-2] -0.945y[n-3]
(V) y[n] = x[n] +1.05y[n-1]
(VI) y[n] = 0.094x[n] + 0.3759x[n-1] + 0.5639x[n-2] + 0.3759x[n-3] + ...
0.094x[n-4] + 0.486y[n-2] + 0.0177y[n-4]
(VII) b = [0.6066,0,2.4264,0,3.6396,0,2.4264,0,0.6066];
a = [1,0,3.1004,0,3.7156,0,2.0314,0,0.4332]
4. For the following different functions, compute the DTFT, then plot the function on a first subplot
and its DTFT magnitude on a second subplot. The sinc function can be evaluated using the function
 1.9. EXERCISES 21
sinc(x). What kind of filter impulse response results from the functions below? What effect does the
parameter c have, and what effect does the length of the vector n have?
(a) n =-9:1:9; c = 0.8; y = sinc(c*n);
(b) n =-9:1:9; c = 0.4; y = sinc(c*n);
(c) n =-9:1:9; c = 0.2; y = sinc(c*n);
(d) n =-9:1:9; c = 0.1; y = sinc(c*n);
(e) n =-39:1:39; c = 0.8; y = sinc(c*n);
(f ) n =-39:1:39; c = 0.4; y = sinc(c*n);
(g) n =-39:1:39; c = 0.2; y = sinc(c*n);
(h) n =-39:1:39; c = 0.1; y = sinc(c*n);
(i) n =-139:1:139; c = 0.8.; y = sinc(c*n);
(j) n =-139:1:139; c = 0.4.; y = sinc(c*n);
(k) n =-139:1:139; c = 0.2.; y = sinc(c*n);
(l) n =-139:1:139; c = 0.1.; y = sinc(c*n);
5. Compute and plot the magnitude and phase of the DTFT of the following impulse responses x. Note
the effect of the frequency parameter f and the difference in impulse responses and especially phase
responses between (a)-(e) and (f )-(j).
(a) n = -10:1:10; f = 0.05; x = ((0.9).ˆ(abs(n))).*cos(f*pi*n)
(b) n = -10:1:10; f = 0.1; x = ((0.9).ˆ(abs(n))).*cos(f*pi*n)
(c) n = -10:1:10; f = 0.2; x = ((0.9).ˆ(abs(n))).*cos(f*pi*n)
(d) n = -10:1:10; f = 0.4; x = ((0.9).ˆ(abs(n))).*cos(f*pi*n)
(e) n = -10:1:10; f = 0.8; x = ((0.9).ˆ(abs(n))).*cos(f*pi*n)
(f ) n = 0:1:20; f = 0.05; x = ((0.9).ˆ(abs(n))).*cos(f*pi*n)
(g) n = 0:1:20; f = 0.1; x = ((0.9).ˆ(abs(n))).*cos(f*pi*n)
(h) n = 0:1:20; f = 0.2; x = ((0.9).ˆ(abs(n))).*cos(f*pi*n)
(i) n = 0:1:20; f = 0.4; x = ((0.9).ˆ(abs(n))).*cos(f*pi*n)
(j) n = 0:1:20; f = 0.8; x = ((0.9).ˆ(abs(n))).*cos(f*pi*n)
6. An ideal lowpass filter should have unity gain for |ω| ≤ ωc and zero gain for |ω| > ωc , and a linear
phase factor or constant sample delay equal to e−j ωM where M represents the number of samples of delay.
Use the inverse DTFT to determine the impulse response that corresponds to this frequency specification.
That is to say, perform the following integration
 π
1
x[n] = X(ej ω )ej wn dω
2π −π
where

1 · e−j ωM |ω| ≤ ωc
X(e ) =
jω
0 |ω| > ωc
After performing the integration and obtaining an expression for x[n], use n = -20:1:20 and M =
10, and obtain four impulse responses corresponding to the following values of ωc :
(a) 0.1π
(b) 0.25π
(c) 0.5π
(d) 0.75π
22 CHAPTER 1. THE DISCRETE TIME FOURIER TRANSFORM
Convolve each of the four resulting impulse responses with a chirp of length 1000 samples and
frequency varying from 0 to 500 Hz. For each of the four impulse responses, plot the impulse response
on one subplot and the chirp response on a second subplot. On a third subplot, plot the result from
numerically computing the DTFT of each impulse response, or instead use the script LV_DTFT_Basic
or the similar short version presented in the text to evaluate the DTFT and plot it in a separate window.
To display on the third subplot, create a script

d = LV xDT F T _Basic(x, M, R)
based on LV_DTFT_Basic, but which delivers the DTFT as the output argument d and which does not
itself create a display.
7. For each of the four impulse responses computed above, obtain a new impulse response by subtracting
the given impulse response from the vector [zeros(1,20),1,zeros(1,20)] (this assumes that the vector n in
the previous example ran from -20 to +20).
Determine what kind of filter the four new impulse responses form. To do this, plot the magnitude
of the DTFT of each original impulse response next to the DTFT of each corresponding new impulse
response.
8. Create a script in accordance with the following call syntax, which should create plots similar to that
shown in Fig. 1.3; test it with the sample calls given below.
function LVxDTFT(x,n,M,R,FreqOpt,FigNo)
% Computes and displays the magnitude, phase, real, and
% imaginary parts of the DTFT of the sequence x having
% time indices n, evaluated over M samples.
% Pass R as 1 to evaluate over pi radians, or 2 to evaluate
% over 2*pi radians
% Use FreqOpt = 1 for symmetrical frequency evaluation
% about frequency 0 or FreqOpt = 2 for an asymmetrical
% frequency evaluation
% Sample calls:
% LVxDTFT([cos(2*pi*25*(0:1:99)/100)],[0:1:99],500,2,1,88)
% LVxDTFT([cos(2*pi*25*(-50:1:50)/100)],[-50:1:50],500,2,1,88)
% LVxDTFT([cos(2*pi*5*(0:1:20)/20)],[0:1:20],100,2,1,88)
% LVxDTFT([cos(2*pi*5*(-10:1:10)/20)],[-10:1:10],100,2,1,88)
% LVxDTFT([cos(2*pi*25*(0:1:100)/100)],[-50:1:50],500,2,1,88)
% LVxDTFT([exp(j*2*pi*25*(0:1:99)/100)],[0:1:99],500,2,1,88)
% LVxDTFT([cos(2*pi*25*(0:1:99)/100)],[0:1:99],1000,2,1,88)
9. Create a script that conforms to the following call syntax, which should create plots similar to that
shown in Fig. 1.6; test it with the sample calls given below.
function LVxDTFT_MS(x,SampOffset,FreqOffsetExp,...
% M,R,TimeOpt,FreqOpt)
% Computes and displays the magnitude, phase, real, and
% imaginary parts of the DTFT of the sequence x, evaluated
% over M samples, then computes the same for a
% modified version of x that has been shifted by
% SampOffset samples and multiplied by a complex
% exponential FreqOffsetExp.
 1.9. EXERCISES 23
% Pass R as 1 to evaluate from 0 to pi radians, or
% Pass R as 2 to evaluate from 0 to 2*pi radians
% Pass TimeOpt as 1 to let n (the time indices for x1
% and x2) be computed as n = -(N-1)/2:1:(N-1)/2 for N odd
% or n = -N/2+1:1:N/2; for even N.
% Pass FreqOpt as 1 for symmetrical frequency computation
% and display (-R*pi to +R*pi, for example) or as 2 for
% frequency comp.and display from 0 to R*pi
% Sample calls:
% LVxDTFT_MS([cos(2*pi*25*(0:1:100)/100)],0,...
% exp(j*2*pi*10*(0:1:100)/100),500,2,1,1)
% LVxDTFT_MS([cos(2*pi*25*(0:1:100)/100)],0,...
% exp(-j*2*pi*10*(0:1:100)/100),500,2,1,1)
% LVxDTFT_MS([cos(2*pi*25*(0:1:100)/100)],0,...
% exp(-j*pi/2),500,2,1,1) % shifts phase
% LVxDTFT_MS([exp(j*2*pi*25*(0:1:100)/100)],0,...
% exp(-j*2*pi*10*(0:1:100)/100),500,2,1,1)
% LVxDTFT_MS([cos(2*pi*25*(0:1:100)/100)],0,...
% exp(j*2*pi*12.5*(0:1:100)/100),1000,2,1,1)
% LVxDTFT_MS([1 0 1],2,1,300,2,1,1)
% LVxDTFT_MS([1 0 1],0,exp(j*2*pi*1*(0:1:2)/3),...
% 100,2,1,1)
% LVxDTFT_MS([cos(2*pi*25*(0:1:100)/100)],0,...
% exp(j*2*pi*12.5*(0:1:100)/100),200,2,1,1)
% LVxDTFT_MS([1 0 1],0,exp(j*2*pi*(0:1:2)/3),300,2,1,1)
10. Compute 1200 evenly distributed frequency samples of the DTFT (i.e., a numerical approximation
of the DTFT) for the three following sequences:
(a) [0.1,0.7,1,0.7,0.1]
(b) [1,0,0,0,1]
(c) [0.95n - 0.85n ] for 0 ≤ n ≤ ∞
11. Numerically compute the inverse DTFTs of the three numeric DTFTs computed in the previous
exercise (for item (c), compute the inverse DTFT for n = 0:1:2).
12. Compute the net output signal obtained by convolving the signal

x = 0.5 cos(2π n/16) + 0.3 sin(5π n/16)

with the impulse response [1,0,-1] two ways, first, by performing the time domain convolution, and second,
by computing the DTFT of the impulse response to determine the magnitude and phase responses for the
frequencies in the signal, and then scaling and shifting the two signal components, and finally summing
the two scaled and shifted signal components to obtain the net output response. Verify that the results
are the same during steady state.
 25

CHAPTER 2

The z-Transform
2.1 OVERVIEW
In the previous chapter, we took a brief look at the Fourier and Laplace families of transforms, and a
more detailed look at the DTFT, which is a member of the Fourier family which receives a discrete
time sequence as input and produces an expression for the continuous frequency response of the discrete
time sequence. With this chapter, we take up the z-transform, which uses correlators having magnitudes
which can grow, decay, or remain constant over time. It may be characterized as a discrete-time variant
of the Laplace Transform. The z-transform can not only be used to determine the frequency response
of an LTI system (i.e., the LTI system’s response to unity-amplitude correlators), it reveals the locations
of poles and zeros of the system’s transfer function, information which is essential to characterize and
understand such systems. The z-transform is an indispensable transform in the discrete signal processing
toolbox, and is virtually omnipresent in DSP literature. Thus, it is essential that the reader gain a good
understanding of it.
The z-transform mathematically characterizes the relationship between the input and output
sequences of an LTI system using the generalized complex variable z, which, as we have already seen, can
be used to represent signals in the form of complex exponentials. Many benefits accrue from this:

• An LTI system is conveniently and compactly represented by an algebraic expression in the vari-
able z; this expression, in general, takes the form of the ratio of two polynomials, the numerator
representing the FIR portion of the LTI system, and the denominator representing the IIR portion.

• Values of z having magnitude 1.0, which are said to “lie on the unit circle” can be used to evaluate
the z-transform and provide a frequency response equivalent to the DTFT.

• Useful information about a digital system can be deduced from its z-transform, such as location of
system poles and zeros.

• Difference equations representing the LTI system can be constructed directly from inspection of
the z-transform.

• An LTI system’s impulse response can be obtained by use of the Inverse z-transform, or by con-
structing a digital filter or difference equation directly from the z-transform, and processing a unit
impulse.

• The z-transform of an LTI system has, in general, properties similar or analogous to various other
frequency domain transforms such the DFT, Laplace Transform, etc.

By the end of this chapter, the reader will have gained a practical knowledge of the z-transform, and
should be able to navigate among difference equations, direct-form, cascade, and parallel filter topologies,
and the z-transform in polynomial or factored form, converting any one representation to another. Addi-
tionally, an understanding will have been acquired of the inverse z-transform, and use of the z-transform
to evaluate frequency response of various LTI systems such as the FIR and the IIR.
26 CHAPTER 2. THE Z-TRANSFORM

2.2 SOFTWARE FOR USE WITH THIS BOOK

The software files needed for use with this book (consisting of m-code (.m) files, VI files (.vi), and related
support files) are available for download from the following website:

http://www.morganclaypool.com/page/isen
The entire software package should be stored in a single folder on the user’s computer, and the full
file name of the folder must be placed on the MATLAB or LabVIEW search path in accordance with the
instructions provided by the respective software vendor (in case you have encountered this notice before,
which is repeated for convenience in each chapter of the book, the software download only needs to be
done once, as files for the entire series of four volumes are all contained in the one downloadable folder).
See Appendix A for more information.

2.3 DEFINITION & PROPERTIES

2.3.1 THE Z-TRANSFORM
The z-transform of a sequence x[n] is:
∞

X(z) = x[n]z−n
n=−∞

where z represents a complex number. The transform does not converge for all values of z; the region of
the complex plane in which the transform converges is called the Region of Convergence (ROC), and
is discussed below in detail. The sequence z−n is a complex correlator generated as a power sequence of
the complex number z and thus X(z) is the correlation (CZL) between the signal x[n] and a complex
exponential the normalized frequency and magnitude variation over time of which are determined by the
angle and magnitude of z.

2.3.2 THE INVERSE Z-TRANSFORM

The formal definition of the inverse z-transform is

1
x[n] = X(z)zn−1 dz (2.1)
2πj
where the contour of integration is a closed counterclockwise path in the complex plane that surrounds
the origin (z = 0) and lies in the ROC.
There are actually many methods of converting a z-transform expression into a time domain
expression or sequence. These methods, including the use of Eq. (2.1), will be explored later in the
chapter.

2.3.3 CONVERGENCE CRITERIA

Infinite Length Causal (Positive-time) Sequence
When x[n] is infinite in length, and identically 0 for n < 0, the ratio of x[n] to zn (or in other words,
x[n]z−n ) must generally decrease in magnitude geometrically as n increases for convergence to a finite
sum to occur.
 2.3. DEFINITION & PROPERTIES 27
Note that if the sequence x[n] is a geometrically convergent series, then the z-transform will also
converge provided that  
 x[n + 1] 
 
 x[n]  < |z|

In terms of numbers, if  
 x[n + 1] 
 
 x[n]  = 0.9

for example, then it is required that |z| > 0.9 for convergence to occur.
If in fact x[n] is a geometrically convergent series, and z is properly chosen, the sum of the infinite
series of numbers consisting of x[n]z−n may conveniently be written in a simple algebraically closed form.

Example 2.1. Determine the z-transform for a single pole IIR with a real pole p having a magnitude
less than 1.0.

The impulse response of such a filter may be written as

[p 0 (= 1), p1 , p2 , p3 , ... pn ]

etc., or to pick a concrete example with the pole at 0.9,

0.9n = [1, 0.9, 0.81, 0.729, ...]

and the z-transform would therefore be:

A(z) = 1 + 0.9z−1 + 0.81z−2 + 0.729z−3 + . . . pn z−n

or in generic terms as

A(z) = 1 + pz−1 + p 2 z−2 + p 3 z−3 + . . . pn z−n

The summation of an infinite number of terms of the form cn where |c| < 1 with

0≤n<∞
is
1
(2.2)
1−c
For the single pole IIR with a pole at p, and by letting c = pz−1 (note that p 0 = 1) in Eq. (2.2),
we get the closed-form z-transform as
1 z
A(z) = −1
= (2.3)
1 − pz z−p
The z-transform in this case is defined or has a finite value for all z with |z| > |p| or
 
p
 <1
z
28 CHAPTER 2. THE Z-TRANSFORM
Example 2.2. Plot the ROC in the z-plane for the z-transform corresponding to the sequence 0.8n u[n].

The z-transform is
1
X(z) =
1 − 0.8z−1
which converges for |z| > 0.8. The shaded area of Fig. 2.1 shows the ROC.

1
0.8
0.6
0.4
Imaginary

0.2
0
−0.2
−0.4
−0.6
−0.8
−1

−1 −0.5 0 0.5 1
Real

Figure 2.1: The Region of Convergence for a system having one pole at 0.8 is shown as the shaded area
outside (not including) a circle of radius 0.8. The unit circle is shown as a dotted circle at radius 1.0.

Example 2.3. Write the z-transform for the sequence 0.7n u[n], evaluate it at 500 values of z lying on
the unit circle (i.e., having a magnitude of 1.0, and radian frequencies between 0 and 2π ), and plot the
magnitude of the result.

Values of z on the unit circle have magnitude greater than 0.7 and thus lie in the ROC for the
z-transform, which is 1/(1- 0.7z−1 ); Figure 2.2 shows the result of running the following m-code, which
computes and plots the magnitude of the z-transform:
radFreq = [0:2*pi/499:2*pi]; z = exp(j*radFreq);
 2.3. DEFINITION & PROPERTIES 29

3.5

2.5
Magnitude

1.5

0.5
0 0.5 1 1.5 2
Frequency, Units of π

Figure 2.2: The magnitude of H (z) = 1/(1 − 0.7z−1 ) evaluated at z = exp(jω) for ω = 0 to 2π .

Zxform = 1./(1-0.7*z.ˆ(-1)); plot(radFreq/pi,abs(Zxform))

For the positive-time or causal sequence x[n], convergence of the z-transform is guaranteed for all
z with magnitude greater than the pole having greatest magnitude (sometimes referred to as the dominant
pole) in a transfer function.
For example, a transfer function having two poles at 0.9 and 0.8, respectively, would have the
z-transform

1
A(z) =
(1 − 0.9z−1 )(1 − 0.8z−1 )
which would converge for all z having |z| > 0.9.
The z-transform is undefined where it does not converge. Consider an example where p = 0.9,
and we choose to evaluate the transform at z = 0.8. We write the first few terms of the definition of the
z-transform as

A(z) = p0 z0 + pz−1 + p 2 z−2 + p 3 z−3 + . . . pn z−n

which in concrete terms would be

0.9 0 0.9 1 0.81 2 0.9 3 0.9 n

A(z) = ( ) +( ) +( ) +( ) + ...( )
0.8 0.8 0.64 0.8 0.8
which clearly diverges, and thus the transform is undefined.
 Another Random Document on
Scribd Without Any Related Topics
 containing a part of this work or any other work associated with
Project Gutenberg™.

1.E.5. Do not copy, display, perform, distribute or redistribute

this electronic work, or any part of this electronic work, without
prominently displaying the sentence set forth in paragraph 1.E.1
with active links or immediate access to the full terms of the
Project Gutenberg™ License.

1.E.6. You may convert to and distribute this work in any binary,
compressed, marked up, nonproprietary or proprietary form,
including any word processing or hypertext form. However, if
you provide access to or distribute copies of a Project
Gutenberg™ work in a format other than “Plain Vanilla ASCII” or
other format used in the official version posted on the official
Project Gutenberg™ website (www.gutenberg.org), you must,
at no additional cost, fee or expense to the user, provide a copy,
a means of exporting a copy, or a means of obtaining a copy
upon request, of the work in its original “Plain Vanilla ASCII” or
other form. Any alternate format must include the full Project
Gutenberg™ License as specified in paragraph 1.E.1.

1.E.7. Do not charge a fee for access to, viewing, displaying,

performing, copying or distributing any Project Gutenberg™
works unless you comply with paragraph 1.E.8 or 1.E.9.

1.E.8. You may charge a reasonable fee for copies of or

providing access to or distributing Project Gutenberg™
electronic works provided that:

• You pay a royalty fee of 20% of the gross profits you derive
from the use of Project Gutenberg™ works calculated using the
method you already use to calculate your applicable taxes. The
fee is owed to the owner of the Project Gutenberg™ trademark,
but he has agreed to donate royalties under this paragraph to
the Project Gutenberg Literary Archive Foundation. Royalty
 payments must be paid within 60 days following each date on
which you prepare (or are legally required to prepare) your
periodic tax returns. Royalty payments should be clearly marked
as such and sent to the Project Gutenberg Literary Archive
Foundation at the address specified in Section 4, “Information
about donations to the Project Gutenberg Literary Archive
Foundation.”

• You provide a full refund of any money paid by a user who

notifies you in writing (or by e-mail) within 30 days of receipt
that s/he does not agree to the terms of the full Project
Gutenberg™ License. You must require such a user to return or
destroy all copies of the works possessed in a physical medium
and discontinue all use of and all access to other copies of
Project Gutenberg™ works.

• You provide, in accordance with paragraph 1.F.3, a full refund of

any money paid for a work or a replacement copy, if a defect in
the electronic work is discovered and reported to you within 90
days of receipt of the work.

• You comply with all other terms of this agreement for free
distribution of Project Gutenberg™ works.

1.E.9. If you wish to charge a fee or distribute a Project

Gutenberg™ electronic work or group of works on different
terms than are set forth in this agreement, you must obtain
permission in writing from the Project Gutenberg Literary
Archive Foundation, the manager of the Project Gutenberg™
trademark. Contact the Foundation as set forth in Section 3
below.

1.F.

1.F.1. Project Gutenberg volunteers and employees expend

considerable effort to identify, do copyright research on,
transcribe and proofread works not protected by U.S. copyright
law in creating the Project Gutenberg™ collection. Despite these
efforts, Project Gutenberg™ electronic works, and the medium
on which they may be stored, may contain “Defects,” such as,
but not limited to, incomplete, inaccurate or corrupt data,
transcription errors, a copyright or other intellectual property
infringement, a defective or damaged disk or other medium, a
computer virus, or computer codes that damage or cannot be
read by your equipment.

1.F.2. LIMITED WARRANTY, DISCLAIMER OF DAMAGES - Except

for the “Right of Replacement or Refund” described in
paragraph 1.F.3, the Project Gutenberg Literary Archive
Foundation, the owner of the Project Gutenberg™ trademark,
and any other party distributing a Project Gutenberg™ electronic
work under this agreement, disclaim all liability to you for
damages, costs and expenses, including legal fees. YOU AGREE
THAT YOU HAVE NO REMEDIES FOR NEGLIGENCE, STRICT
LIABILITY, BREACH OF WARRANTY OR BREACH OF CONTRACT
EXCEPT THOSE PROVIDED IN PARAGRAPH 1.F.3. YOU AGREE
THAT THE FOUNDATION, THE TRADEMARK OWNER, AND ANY
DISTRIBUTOR UNDER THIS AGREEMENT WILL NOT BE LIABLE
TO YOU FOR ACTUAL, DIRECT, INDIRECT, CONSEQUENTIAL,
PUNITIVE OR INCIDENTAL DAMAGES EVEN IF YOU GIVE
NOTICE OF THE POSSIBILITY OF SUCH DAMAGE.

1.F.3. LIMITED RIGHT OF REPLACEMENT OR REFUND - If you

discover a defect in this electronic work within 90 days of
receiving it, you can receive a refund of the money (if any) you
paid for it by sending a written explanation to the person you
received the work from. If you received the work on a physical
medium, you must return the medium with your written
explanation. The person or entity that provided you with the
defective work may elect to provide a replacement copy in lieu
of a refund. If you received the work electronically, the person
or entity providing it to you may choose to give you a second
opportunity to receive the work electronically in lieu of a refund.
If the second copy is also defective, you may demand a refund
in writing without further opportunities to fix the problem.

1.F.4. Except for the limited right of replacement or refund set

forth in paragraph 1.F.3, this work is provided to you ‘AS-IS’,
WITH NO OTHER WARRANTIES OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO WARRANTIES OF
MERCHANTABILITY OR FITNESS FOR ANY PURPOSE.

1.F.5. Some states do not allow disclaimers of certain implied

warranties or the exclusion or limitation of certain types of
damages. If any disclaimer or limitation set forth in this
agreement violates the law of the state applicable to this
agreement, the agreement shall be interpreted to make the
maximum disclaimer or limitation permitted by the applicable
state law. The invalidity or unenforceability of any provision of
this agreement shall not void the remaining provisions.

1.F.6. INDEMNITY - You agree to indemnify and hold the

Foundation, the trademark owner, any agent or employee of the
Foundation, anyone providing copies of Project Gutenberg™
electronic works in accordance with this agreement, and any
volunteers associated with the production, promotion and
distribution of Project Gutenberg™ electronic works, harmless
from all liability, costs and expenses, including legal fees, that
arise directly or indirectly from any of the following which you
do or cause to occur: (a) distribution of this or any Project
Gutenberg™ work, (b) alteration, modification, or additions or
deletions to any Project Gutenberg™ work, and (c) any Defect
you cause.

Section 2. Information about the Mission

of Project Gutenberg™
Project Gutenberg™ is synonymous with the free distribution of
electronic works in formats readable by the widest variety of
computers including obsolete, old, middle-aged and new
computers. It exists because of the efforts of hundreds of
volunteers and donations from people in all walks of life.

Volunteers and financial support to provide volunteers with the

assistance they need are critical to reaching Project
Gutenberg™’s goals and ensuring that the Project Gutenberg™
collection will remain freely available for generations to come. In
2001, the Project Gutenberg Literary Archive Foundation was
created to provide a secure and permanent future for Project
Gutenberg™ and future generations. To learn more about the
Project Gutenberg Literary Archive Foundation and how your
efforts and donations can help, see Sections 3 and 4 and the
Foundation information page at www.gutenberg.org.

Section 3. Information about the Project

Gutenberg Literary Archive Foundation
The Project Gutenberg Literary Archive Foundation is a non-
profit 501(c)(3) educational corporation organized under the
laws of the state of Mississippi and granted tax exempt status
by the Internal Revenue Service. The Foundation’s EIN or
federal tax identification number is 64-6221541. Contributions
to the Project Gutenberg Literary Archive Foundation are tax
deductible to the full extent permitted by U.S. federal laws and
your state’s laws.

The Foundation’s business office is located at 809 North 1500

West, Salt Lake City, UT 84116, (801) 596-1887. Email contact
links and up to date contact information can be found at the
Foundation’s website and official page at
www.gutenberg.org/contact
 Section 4. Information about Donations to
the Project Gutenberg Literary Archive
Foundation
Project Gutenberg™ depends upon and cannot survive without
widespread public support and donations to carry out its mission
of increasing the number of public domain and licensed works
that can be freely distributed in machine-readable form
accessible by the widest array of equipment including outdated
equipment. Many small donations ($1 to $5,000) are particularly
important to maintaining tax exempt status with the IRS.

The Foundation is committed to complying with the laws

regulating charities and charitable donations in all 50 states of
the United States. Compliance requirements are not uniform
and it takes a considerable effort, much paperwork and many
fees to meet and keep up with these requirements. We do not
solicit donations in locations where we have not received written
confirmation of compliance. To SEND DONATIONS or determine
the status of compliance for any particular state visit
www.gutenberg.org/donate.

While we cannot and do not solicit contributions from states

where we have not met the solicitation requirements, we know
of no prohibition against accepting unsolicited donations from
donors in such states who approach us with offers to donate.

International donations are gratefully accepted, but we cannot

make any statements concerning tax treatment of donations
received from outside the United States. U.S. laws alone swamp
our small staff.

Please check the Project Gutenberg web pages for current

donation methods and addresses. Donations are accepted in a
number of other ways including checks, online payments and
credit card donations. To donate, please visit:
www.gutenberg.org/donate.

Section 5. General Information About

Project Gutenberg™ electronic works
Professor Michael S. Hart was the originator of the Project
Gutenberg™ concept of a library of electronic works that could
be freely shared with anyone. For forty years, he produced and
distributed Project Gutenberg™ eBooks with only a loose
network of volunteer support.

Project Gutenberg™ eBooks are often created from several

printed editions, all of which are confirmed as not protected by
copyright in the U.S. unless a copyright notice is included. Thus,
we do not necessarily keep eBooks in compliance with any
particular paper edition.

Most people start at our website which has the main PG search
facility: www.gutenberg.org.

This website includes information about Project Gutenberg™,

including how to make donations to the Project Gutenberg
Literary Archive Foundation, how to help produce our new
eBooks, and how to subscribe to our email newsletter to hear
about new eBooks.
Welcome to Our Bookstore - The Ultimate Destination for Book Lovers
Are you passionate about books and eager to explore new worlds of
knowledge? At our website, we offer a vast collection of books that
cater to every interest and age group. From classic literature to
specialized publications, self-help books, and children’s stories, we
have it all! Each book is a gateway to new adventures, helping you
expand your knowledge and nourish your soul
Experience Convenient and Enjoyable Book Shopping Our website is more
than just an online bookstore—it’s a bridge connecting readers to the
timeless values of culture and wisdom. With a sleek and user-friendly
interface and a smart search system, you can find your favorite books
quickly and easily. Enjoy special promotions, fast home delivery, and
a seamless shopping experience that saves you time and enhances your
love for reading.
Let us accompany you on the journey of exploring knowledge and
personal growth!

ebookgate.com