Implementation __________________________
AES Survey: Specification
Hudson C. Stansbury
George Mason University
October 19, 2004
__________________________
Introduction
The purpose of this project will be to compare various existing software
implementations of the Advanced Encryption Standard (AES) algorithm
primarily based on performance. AES is still young next to its
predecessor, the Data Encryption Standard (DES), and its security and
performance qualities will determine its future as the new predominate
standard block cipher algorithm. Two most important implementations
currently available are the reference implementation provided at
http://www.esat.kuleuven.ac.be/~rijmen/rijndael/rijndaelref.zip, and the
optimized implementation, provided at
http://www.esat.kuleuven.ac.be/~rijmen/rijndael/rijndael-fst-3.0.zip.
__________________________
System Characteristics and Additional Software Required
The experiments will be performed on Windows XP over a 1.79 GHz Pentium
processor, as well as a GMU-owned SunOS system, cpeo2, running on a 400
MHz Sparc processor. Each implementation might be available as compiled
code, assembly, or source code in a high level language (e.g. C++, Java),
the latter two requiring their own assemblers and compilers. We intend to
diversify these compilers used for the experiments. Among other
compilers, we plan to use the following:
1.Java 2 SDK, Standard Edition Version 1.4.1
2.ActivePerl Interpreter Version 5.8.0
3.C and C++ compilers from GCC and from Sun's Forte Developer
The C and C++ compilers to be used on the Windows system have not yet been
selected.
It may be difficult to separately examine the effects of hardware and
operating system because only two machines will be used.
__________________________
Input/Output Format
The input will consist of plaintext files for encryptions, ciphertext
files for decryptions, and keys provided as standard input. Output will
consist of the ciphertext files for encryptions, plaintext files for
decryptions, and performance characteristics, including the amount of data
encrypted. Different implementations may have different interfaces based
on whether the program interacts with a file system or standard I/O or
some other interface.
__________________________
Function Description
AES is an iterative shared-key block cipher algorithm, which allows for
key sizes and block sizes of 128 bits, 192 bits, and 256 bits. The
original specification of the AES algorithm can be found at
http://csrc.nist.gov/publications/fips/fips197/fips-197.pdf.
__________________________
Plan of Experiments
We will perform experiments with several different implementations of AES
and compare them based on the following qualities:
1.author
2.language
3.compiler
4.platform
5.block size (128, 192, or 256 bits)
6.key size (128, 192, or 256 bits)
7.implementation structure
In each experiment, one implementation of AES will be used to encrypt or
decrypt as much data as possible in certain interval. time The input
plaintexts will each come from a file of randomly-generated data, and the
key will be selected from a list of randomly-generated keys.
__________________________
Schedule
Monday November 8:Gain preliminary understanding of the AES
algorithm and its mathematical basis.
Monday November 15:Gain preliminary understanding of implementation
decisions involved in AES and the range of existing implementations of
AES.
Monday November 15:Download, compile, and run the various
implementations. Perform any necessary configurations and informally
experiment with the programs in order that they will be most effectively
used for formal experimentation.
Wednesday November 17:Gain a closer theoretical understanding of the
qualities in the implementation of AES that can affect its performance.
Wednesday December 1:Devise a precise set of inputs to be used for the
AES algorithm, document the justification for using those inputs, and
perform the experiments. These steps will be done iteratively in order to
use the results of some experiments to design the inputs of other
experiments.
Monday December 6:Document results and conclusions of the
experiments. Describe how different implementations of AES and,
generally, how different qualities of those implementations might be more
or less useful in various scenarios.
Monday December 13:Finalize all documentation.
__________________________
Possible Specification Changes
The range of compilers and languages used on each system may evolve,
especially since some key compilers have not yet been specified.
Some of the seven currently-specified comparison criteria (see plan of
experiments) may be eliminated. Also the criterion called implementation
structure is expected to eventually describe more ...