Re: Minimum memory footprint implementations

From: Zbigniew Laskarzewski (zbygumil_at_interia.pl)
Date: 05/27/03 

Date: 27 May 2003 12:04:29 -0700

I have a similar problem. I have to implement encryption/decryption
and authentication on 8051, where are also other programs. I have few
memory for code and RAM and EEPROM. I'm going to implement 2
algorithms, which use 1-way hash functions. To autentication - HMAC
(NIST standard), to encryption/decryption - MDC desined by Peter
Gutmann or Luby-Rakoff. But I didn't find papers about MDC or
Luby-Rakoff.

Did anybody know where can I find papers about Gutmann's MDC and
Luby-Rakoff(except Schneier)?

HMAC - http://csrc.nist.gov/publications/fips/fips198/fips-198a.pdf

I implemented SHA-1 on 8051 ( ~1500B code, ~70B RAM ). This
implementation of 1-way hash function will used to authentication and
encryption/decryption.
HMAC and MDC or Luby-Rakoff use not lot of memory for code and RAM
(except implementation 1-way hash function).
Key to authentication and encryption/decryption have 20B (EEPROM).
Finding a hash from one 64-byte block of data == 40 miliseconds
What do you think about this conception?
Any suggestions?
(Sorry for my english)

Zbyszek