Comparing candidates
for the new Advanced Encryption Standard
using reconfigurable hardware

Kris Gaj and Pawel Chodowiec

George Mason University

http://ece.gmu.edu/crypto-text.htm


Click here to start


Table of Contents

Comparing candidates for the new Advanced Encryption Standard using reconfigurable hardware

Outline

Current American encryption standard

Controversies surrounding DES

Deep Crack

Why a new standard?

Why a contest?

Rules of the contest

AES contest - First Round

AES: Candidate algorithms

First round June 1998 - August 1999

Survey filled by 104 participants of the Second AES Conference in Rome, March 1999

AES Finalists (1)

AES Finalists (2)

Second round August 1999 - August 2000

AES Contest: Second Round

Primary ways of implementing cryptography in hardware

Advantages of using FPGAs for comparison of the AES cadidates

Why to use FPGAs in a final product ?

Capabilities of reconfiguration (1)

Capabilities of reconfiguration (2)

Target FPGA devices: High Performance

Target FPGA devices: Low Cost

Basic building blocks of FPGA devices

Methodology and Tools

Top level block diagram

Basic architecture

k-rounds Loop Unrolling

Loop Unrolling: Speed vs. Area

k-stage Outer-Round Pipelining

Pipelined operation of the encryption unit

Outer-Round Pipelining: Speed vs. Area

Non-feedback cipher modes - ECB

Feedback cipher modes - CBC

k-stage Inner-Round Pipelinig

Inner-Round Pipelining: Speed vs. Area

Resource Sharing

Examples of functions F that can be shared

Performance of alternative archtectures: in non-feedback cipher modes (ECB, counter)

Performance of alternative archtectures: in feedback cipher modes (CBC,CFB,OFB)

Basic architecture: Speed, XC 4000XL

Basic architecture: Area, XC 4000XL

Basic architecture: Speed, Virtex

Basic architecture: Area, Virtex

Basic architecture: Speed/Area, Virtex

Encryption in cipher feedback modes (CBC,CFB,OFB)

Comparison with results of other groups: Speed, Virtex FPGA

Comparison with results of other groups: Area, Virtex FPGA

Comparison with results of the NSA group (1)

Our basic architecture of Serpent

NSA's basic architecture of Serpent

Comparison with results of the NSA group (2)

Comparison with results of other groups (1)

Comparison with results of other groups (2)

Target FPGA devices: Medium Performace

Basic architecture: Speed, Altera 10K250A

Basic architecture: Area, Altera 10K250A

Conclusions (1)

Major operations of AES finalists

Critical path: Time

Inner-Round Pipelining

Choosing optimum architecture for non-feedback cipher modes

Full Mixed Inner and Outer-Round Pipelining

Encryption in non-feedback modes (ECB, counter) decryption in all modes [estimations]

NSA architecture for non-feedback cipher modes Full outer-round pipelining

Our architecture for non-feedback cipher modes Full mixed inner- and outer-round pipelining

NSA: Outer-round pipelining: Speed CMOS ASIC

Conclisions (2)

Need for interleaved operating modes

Survey filled by 167 prticipants of the Third AES Conference, April 2000

How NIST is going to make a final decision ?

Security: Theoretical attacks better than exhaustive key search

Security: Authors of attacks

Efficiency in software: NIST-specified platform

Efficiency in software: NIST tests

Efficiency in software: Ranking of encryption speeds for various platforms

Efficiency in software: Key setup

Efficiency in software

Efficiency in software: Conclusions

Flexibility: Criteria

Ranking by participants of the AES3 Conference

Most likely winner(s) (1)

Most likely winner(s) (2)

Most likely winner(s) (3)

Author: Krzysztof Gaj

Email: kgaj@gmu.edu

Home Page: http://ece.gmu.edu/crypto-text.htm