Re: Using SURF as a random data source

On Wed, Mar 24, 2004 at 14:10:44 -0600,
  Bruce Guenter <[email protected]> wrote:
> Greetings.
> 
> Is it reasonable to use surf as a cryptographically strong random data
> source?

I doubt that SURF has received the kind of analysis attention that AES has.
If you don't need speed as much as security, then using AES in counter
mode might be a better choice.

> Obviously, if I use random secret data for both the seed and the input,
> the result will also be secret.  The biggest question I have is how to
> generate more than one output from the same secret in a robust manner.

I think you only need to keep the seed secret. You should be able to
just use a counter for the input. You can't do this with the seeds.
When you need to rekey, you need to generate a new random seed.

> If the math for SURF is sound (and I have no reason to doubt it is),
> then simply treating the input as one long 384-bit counter will produce
> a random number stream that is 2^(384*8) words long.  This seems to be
> the simplest way of producing a random data source with a known period.
> Will this work as described?

I believe so.