[2776] in cryptography@c2.net mail archive
Re: On Mixture of Stream and Block Cipher
daemon@ATHENA.MIT.EDU (Mok-Kong Shen)
Thu May 28 13:10:50 1998
Date: Thu, 28 May 1998 17:18:14 +0100
From: Mok-Kong Shen <mok-kong.shen@stud.uni-muenchen.de>
To: Sunder <sunder@brainlink.com>
CC: cryptography@c2.net
Sunder wrote:
>
> Mok-Kong Shen wrote:
>
> > According to Menezes et al., Handbook of Cryptography, block ciphers
> > are memoryless while stream ciphers have memory and are also called
> > state ciphers. The distinction between the two types is nevertheless
> > not definitive.
>
> If you allow yourself to bend the definitions you'll easily come up with
> ways to generate stream cyphers out of block cyphers and vice versa. Just
> think of a stream cypher as a very small block, and a block cypher as a
> stream of very large characters. i.e. 8 bits versus 64 bits. :)
>
> The problem is what to do with the excess space in a block cypher when you
> don't have data. Sure, you could padd the data, but there comes a question
> of where to pad it and how to prevent the padding from being plain text.
>
> i.e. if your block is 64 bits, and you want to send only one byte, what
> do you do? pad with all zeros? If so, a known plaintext attack could
I am considering proglems of a different scale than that of yours.
I have in mind cases where the amount of information of the message
is not too small (at least a number of records of e.g. 80 characters),
thus justifying putting in length information or beginning/terminating
characters (there can be several), with the rest of the message padded
by random bits. This presupposes of course that the encryption is
sufficiently strong such that such extra information does not aid the
analyst in his work. In fact one could also do without such extra
information for normal messages, since the receiver knows where the
message ends, namely where the words no longer make sense.
M. K. Shen
