Anybody involved in stream ciphering using m-sequence LFSRs??

[DHoran]

I'm currently doing a project involving the generation of pseudo
random numbers using LFSRs (various configurations of m-length
registers). At the moment I'm gathering literature so any suggestions
and references on attacks or configs are welcome!

[Bob Harris]

Do you mean m different LFSRs, combined in some way? Or just a single LFSR
with an m-bit register?

Do you expect the attacker to be able to see a short subsequence of the LFSR
output bits? Or instead only some sequence of cipher text which was derived
by some combination of plaintext and LFSR output?

Bob H

[DHoran]

I mean a number of LFSRs of different lengths and feedback polys
(lengths relatively prime and fb polys primative) combined together in
some way to give a maximal length sequence. In other words a type of
combination generator. Part of my brief is to carry out statistical
testing on the RN produced to establish its quality, and then to try
some attacks.

As to the type of attack, I'll probably look at both ciphertext-only
attack, and an attacker having access to 2n bits of the keystream
(Berlekemp-Massey algorithm). I'll also be looking at correlation
attacks and the whole correlation immunity vs. linear complexity trade
off.

As I said at the moment I'm gathering information and literature so
any suggestions, links or references would be appreciated

Thanks
DH