Scribd
Upload
465 views

Qcrypt: Secure High-Speed Communication Based On Quantum Key Distribution

QCrypt provides high-speed encryption and quantum security. It integrates quantum key distribution (QKD) with terabit encryption schemes. QCrypt aims to improve information security in the 21st century. It works on developing 100Gb/s encryption, removing throughput limits, enhancing security with physically unclonable functions, and winning the SHA-3 hash function contest with the BLAKE candidate. QCrypt has achieved over 1Mbps secret key rates from QKD up to 19km and stable detection rates over 10 hours. It has developed a 40Gb/s and 100Gb/s AES encryption core and optimized the network interface. Research on physically unclonable functions using DRAM and SRAM is underway.

Uploaded by

nanoteraCH
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
465 views

Qcrypt: Secure High-Speed Communication Based On Quantum Key Distribution

QCrypt provides high-speed encryption and quantum security. It integrates quantum key distribution (QKD) with terabit encryption schemes. QCrypt aims to improve information security in the 21st century. It works on developing 100Gb/s encryption, removing throughput limits, enhancing security with physically unclonable functions, and winning the SHA-3 hash function contest with the BLAKE candidate. QCrypt has achieved over 1Mbps secret key rates from QKD up to 19km and stable detection rates over 10 hours. It has developed a 40Gb/s and 100Gb/s AES encryption core and optimized the network interface. Research on physically unclonable functions using DRAM and SRAM is underway.

Uploaded by

nanoteraCH
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 31

QCrypt

RTD 2010

Secure High-Speed Communication based on Quantum Key Distribution


Norbert Felber (for Nicolas Gisin)

Overview

100Gbit/s Encryption & Quantum Security Quantum Key Distribution (QKD) Challenges High-Speed Encryption Challenges Physically Unclonable Functions SHA-3 Hash Functions

Qcrypt +
Interfaces
Quantum Key Distribution

Vision
HES-SO
Terabit Encryption

UNIGE
Terabit Quantum Encryption

ETHZ

Qcrypt is built on Quantum


Industry id Quantique Information Technologies & is integrating this with Terabit Encryption schemes to improve information security in the 21st century

ENGINEERING COMPLEX SYSTEMS FOR HEALTH, SECURITY AND THE ENVIRONMENT

QCrypt:100Gbit/s Encryption & Quantum Security


10111100 00110101 11101010 00110001

40..100Gb/s enCryption and decryption


highest security for the next-generation communication speed

10111100 00110101 11101010 00110001

high-speed Quantum Key Distribution


the impossibility of Q cloning is one of the best established facts in Science

QCrypt System: QKD and Fast EnCryption


10 GE TDM 100 GE AES GHASH Optical Link 100 GE AES GHASH TDM 10 GE GHASH AES GHASH AES 10 GE

10 GE

Key Management & Synchronisation Key

Key Management & Synchronisation Key

QKD: Rapid Gated Single Photon Detectors


Challenges Fast pulse amplitude modulation Rapid gated single photon detectors Stable and high interference visibility Synchronization of Alice and Bob and

Secret key distillation engine

QBER Visibility
0 : 0 , 1 : 0 , 0 1 e
2

QKD: Rapid Gated Single Photon Detectors


130 ps

Peltier cooled InGaAs diode


Gate rate Dead time Afterpulse probability 1.25 GHz 8 ns < 1%

Maximum detection rates ~100 MHz Best performance in academia worldwide!

QKD: Secret Key Distillation


Challenges Fast pulse amplitude modulation Rapid gated single photon detectors Stable and high interference visibility
Sifting Timing and base information

Error estimation

Random sampling for QBER

Error correction

LDPC forward error correction

Error verification

CRC check

Synchronization of Alice and Bob and Secret key distillation engine

Privacy amplification

Toeplitz hashing

Authentication

Polynomial hashing

0 : 0 ,

1 : 0 ,

0 1 e

QKD: Secret Key Distillation


Error correction using LDPC decoder: 802.11n code (Z=81), syndrome encoding Flexible code rates (, , , ) Throughput decrease of 0.5% at 6% QBER
10
0

10

-1

FER

10

-2

OMS, beta = 0.5, {-2 , +2 } MinSum, {-1,+1} Eagle, best choice MinSum, best choice Eagle, 5 bit R, T; 4 bit Q Eagle, 4 bit R; 5 bit T; 4 bit Q Eagle, 3 bit R; 5 bit T; 4 bit Q 0.04 0.06 0.08 0.1 QBER 0.12 0.14 0.16

10

-3

10

-4

0.02

Privacy amplification using Toeplitz matrices: Matrix-vector multiplication: 106 x 105 Random matrix with diagonal structure Flexible compression ratio

QKD: Intermediate Results


10 10
Key rates [s ]
8

0 km

50 km
Sifted rate Error corrected rate Secret rate

100 km

0.10 0.08 0.06 0.04 0.02 0.00


QBER

10 10 10 10

-1

-5

-10 Transmission [dB]

-15

-20

Results
Synchronization and sifting channel running > 1 Mbps estimated secret key rate up to 19 km Stable detection rates over 10 hours (tracking not yet implemented)

Crypt: The Fast Encryption Prototype

Crypt: The Fast Encryption Prototype

Crypt: Troughput Limits


10 GE TDM 100 GE AES GHASH Optical Link 100 GE AES GHASH TDM 10 GE GHASH AES GHASH AES 10 GE

10 GE

Key Management & Synchronisation Key

Key Management & Synchronisation Key

Challenges Alternative 100Gb/s encryption core Removing throughput limits Strong testbench for network interfaces Enhancing system security (PUF) Win SHA-3 Hash function contest

Crypt: Troughput Limits


Problem
Input: 10x10Gb/s gapless packets Encryption produces low overhead (synchronization)

Authentication produces high overhead (MAC)


>100Gb/s data to be transmitted Ethernet handles this by discarding packets and resend requests Throughput tests with Network Analyzer report <100Gb/s^) idQ: severe commercial disadvantage
^) depending on packet lengths

Problem solved
More efficient packet multiplexing method Additional link for supplementary data combined from all 10Gb/s channels

Implemented as 9*x10Gb/s + 1x(10-m)Gb/s, using the


mGb/s for supplementary data (m < 3 typ.) Alternative: separate link (WDM) or service channel of Q-system
*) currently 3 + 1 for 40Gb/s demo

Crypt: Troughput Limits


Changes: adapted significantly reduced removed

Consequences:

88%
24%

52% ALM usage


21% Block RAMs

FPGA Place & Route time 20h 2h

Crypt: System Verilog Test Environment


10 GE TDM 100 GE AES GHASH Optical Link 100 GE AES GHASH TDM 10 GE GHASH AES GHASH AES 10 GE

10 GE

Key Management & Synchronisation Key

Key Management & Synchronisation Key

Challenges Alternative 100Gb/s encryption core Removing throughput limits Strong testbench for network interface debugging Enhancing system security (PUF) Win SHA-3 Hash function contest

Crypt: System Verilog Testbench


Open Verification Methodology (OVM):

Complete verification methodology for large gate-count, IP-based SoCs Transaction channels: classes providing mechanisms to manage data (FIFO, synchronization, etc.)

Generation of random values (data, packet lengths) Virtual links to access nodes inside the DUT Automatic checkers

Crypt: System Verilog Testbench


Open Verification Methodology (OVM):

Example of randomized Ethernet frame

Generation of random values (data) Generation of random packet lengths

QCrypt: System Security?


Crypt: technically secure

Challenges Alternative 100Gb/s encryption core Removing throughput limits Strong testbench for network interfaces

Enhancing system security Win SHA-3 Hash function contest

QKD: provably secure Side channels need to be secured too!

QCrypt: System Security?


10 GE TDM 100 GE AES GHASH Optical Link 100 GE AES GHASH TDM 10 GE GHASH AES GHASH AES 10 GE

10 GE

Key Management & Synchronisation Secret Key

Key Management & Synchronisation Secret Key

PUF: Physically Unclonable Functions


Requirements to a good PUF
is unclonable has a large input data width provides minimal output data width > 48 bits 80 128 bits
1 0 0 1 1 1 0 1 0 0 1 1 1 0 1 0 1 1 1 0 0 0 1 1 0 0

generates random output (for different inputs)


0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 1

PUF
unique physical properties per copy

PUF: Physical Effects


unique physical properties per copy CMOS process variations in VLSI chips
Transistors geometry variations doping variations layer thickness variations . capacitance leakage currents thresholds .

PUF: Variants
PUFs proposed so far:
Race conditions: delay differences in equal pairs of signal paths Ring oscillators: frequency variations Static RAM (SRAM): power-up pattern

Optical: light propagation in passivation layer to on-chip photo diodes

Our proposal: Dynamic RAM PUF (DRAM PUF)


Patent filed: Generating Unique Numbers Using Charge Decay Phenomena
(the patent covers several other charge-decay based effects suitable for PUFs)

PUF: DRAM PUF Principle


data DRAM storage array DRAM cell

write pattern = PUF input (raw) state is stored on capacitors

refresh is disabled
leakage (de)charges capacitors physical variations read word(s) = PUF ouput (raw) sense amplifiers discriminate 0|1 physical variations

observe: all cells in column contribute to leakage!

addr

PUF: DRAM PUF Readout Algorithm


U(t) Udd
U
half way

write pattern

reliable retention

individual responses = zone of interest for PUFs best case true random measured voltages

no correlation with initial conditions

worst case

Uss

typical wait time (t25%)

Initialization: find wait time


repeat write pattern wait t(k++) read pattern until 25% cells toggled

PUF readout: evaluate function


(input pre-processing) write pattern wait t25% read word(s) (output post-processing)

PUF: Comparison
unclonable Race conditions Ring oscillators Optical SRAM DRAM yes yes yes yes yes repetition rate Race conditions Ring oscillators Optical SRAM DRAM high high high low low inputs some some some none many cost, effort average average very high low low outputs few few few many many random (yes) (yes) (yes) no yes^ platform ASIC*, FPGA* ASIC*, FPGA* opto process* ASIC*, FPGA*, Memory* ASIC, Memory * published ^ to be proved

PUF: Applications
Some general applications:

Identification (Challenge Response)


Protected encryption key storage Pseudo random number generator True random number generator (slow) En/Decryption (slow)

Some specific applications for QCrypt:


Q Crypt boxes pairing Protection of Q Crypt quantum key connection Alice Bob boxes pairing Authentication of FPGA bit stream reload (power-up) Protection of FPGA firmware update

SHA-3: Hash Functions


HASH Function BLAKE by Jean-Philippe Aumasson, Luca Henzen, Willi Meier and Raphael C.-W. Phan: Final Round of the NIST SHA-3 Candidate Competition (with Grstl, JH, Keccak, and Skein)
all SHA-3 candidate algorithms ETHZ-IIS: smallest implement. with 2.5Gb/s GMU-CER *): lowest AT product Layout of 65nm CMOS Chip (IIS) 3 SRAMS for PUF investigations
*) ECE Department, George Mason University, Virginia USA {kgaj,mrogawsk,ehomsiri,jpkaps}@gmu.edu

Chip SHABZIGER

Crypt: Intermediate Results


Results
40 & 100Gb/s AES GCM core
Serpent & OCB replacement Optimized network interface with Throughput limit removed FPGA resources freed Strong testbench PUF Research started: Patent filed Test chip with PUF SRAMs ready Test chip with PUF DRAMs in fabrication SHA-3 candidate BLAKE good chances to win contest against Keccak and Skein
DYNAMITE: 4 DRAMs 32x64 for PUFs cell design, MOS caps 180nm CMOS, (1.5mm)2 SHABZIGER: 3 SRAMs for PUFs All SHA-3 candidates 65nm CMOS, (1.9mm)2

Group of Applied Physics (GAP)

Institute of Electrical Engineering (IEL)

REDS HEIG-VD, Yverdon 4 HES-SO Geneva

idQuantique SA Carouge

Integrated Systems Laboratory (IIS)

Left to right: Raphael Houlmann1, Charles Ci Wen Lim1, Patrick Trinkler5, Olivier Auberson3, Rob Thew1, Gilles Curchod3, Jeremy Constantin2, Pascal Junod3, Nino Walenta1, Hugo Zbinden1, Andy Burg2, Norbert Felber6 (Co-PI), Etienne Messerli3 (Co-PI), Natalia Kulesza5, Laurent Monat5, Yoan Graf3, Nicolas Gisin1 (PI), Olivier Guinnard1, Julien Izui4, Christoph Keller6, Fabien Vannel4

Posters
A high speed QKD prototype based on the coherent one-way protocol
by Nino Walenta, presented by Charles Ci Wen Lim

100Giga Fast Encryption Engine


Olivier Auberson, Gilles Curchod, Etienne Messerli

Physically Unclonable Functions for Secure Hardware


Christoph Keller

You might also like