gfgs pedo wrote:
> 
> hi,
> 
> Here are two ideas which came up in my mind.
> Since I have done a few diagrams for illustration and
> thought that it will not be a good idea as
> attachment,I have put the ideas at the following url
> http://www.ircsuper.net/~neo/
> 
> I sincerely appreciate ur comments.Thank u for ur
> time.

Random numbers used in any security application must
meet far stronger requirements than in other applications,
such as simulations etc. The standard reference is:
http://www.ietf.org/rfc/rfc1750.txt

A draft of an updated version is also available:
http://search.ietf.org/internet-drafts/draft-eastlake-randomness2-02.txt

There's also some good material at www.counterpane.com,
in the papers section under "yarrow".

My guess is that neither of your suggestions is useful for
security applications. 

In one case, the input or seed is a text file, not a remarkably
random object, and completely useless if the enemy can discover
or guess what file is used. Also, I cannot tell if your method
is either as secure as or more efficient than a more direct
approach using standard crypto prinitives. For example, one
might just hash the textfile and use the result to key a block
cipher in counter mode.

In the other case, the key seems to be choice of a prime
number. You need a large prime, perhaps 100 bits or more,
to make brute force guessing sufficiently hard. Is your
arithmetic then acceptably efficient? Can an enemy deduce
the prime, or even narrow down the search dangerously,
if he sees enough of your output?

Reply via email to