Re: fixed block size


"Antony Clements" <antony.clements@xxxxxxxxxxxxxxx> wrote in message
news:4630682e$0$11540$afc38c87@xxxxxxxxxxxxxxxxxxxxxxx
most of you here are familiar with one incarnation of my crypto scheme or
another, trivially weak as they are. i have been saving time and effort by
only rewriting the cipher processes, but i just came to a rather stunning
realisation, that if true, would mean i have to start from scratch. in
every rewrite of the cipher process the block size has always been fixed
at 512 bits irrespective of the size of the key. it seems to me with my
limited knowledge of this area, that always having a fixed block size
leads to an exploitable point of attack on the system. given that the
system only allows keys of 128,256,384 and 512 bits, this seems to
severely limit the amount of work needed by an attacker.

the scheme has always been a tri-layer scheme, i've only ever attempted to
describe the layer that i myself have written as it seems rather pointless
to mention the other two layers, but now that i've mentioned that they
exist someone is bound to ask what they are. i'm not going to go into
detail because that will just annoy people with my lack of skill and
knowledge. it should be sufficient to say that the primary layer is my own
code, the nonce for the primary layer is protected with twofish, and then
AES 128 is laid over the top of both. each layer has it's own key. if
someone wants a description of the latest incarnation of the primary layer
feel free to ask, but i doubt anyone will.

given my realisation, on top of every other weakness people have found in
my layer, is a fixed block size in this instance yet another avenue of
attack?


Optimum block size is determined by channel bandwidth and delay through the
channel.
Not by your code.


.