Re: Analysis of /dev/random

New Message Reply About this list Date view Thread view Subject view Author view

David Honig (honig@sprynet.com)
Fri, 09 Apr 1999 11:11:18 -0700


At 10:26 AM 4/9/99 -0400, Ge' Weijers wrote:
>I wrote a simple utility a few months back that generates 'random'
>output. It's basically the TEA block cipher encrypting a counter.

>... but it's not
>useful as a cryptographic random generator ... the output is
>perfectly compressible given the right statistical model :-)

[ie the key]

>It's easy to fool MUST.
>
>Ge'

I agree with everything except the last. MUST is a function.
My use of it as a measure of entropy is valid.

Now, any *extrapolation* that /dev/random is *crypto-secure* depends
on several other things (such as its reliance on noncomputational
(ie, physical) processes, proper hashing, etc.)

I know of no perfect tests for 'randomness' other than
compression-like techniques (e.g., MUST) and structure-finding
techniques (e.g., Diehard). (The latter are similar to running
a monte carlo sim driven by the RNG)

So, if you have a technique that can distinguish between a good
PRNG and a TRNG, *without the key*, I'd like to understand it.

(Because doing so for a given block cipher-based PRNG is equivalent to
analytically breaking the cipher, if the PRNG reveals all the bits of the
block cipher, no?)

......

When I was playing with ways of distilling audio hiss into
quality randomness, I needed a gold standard.
I used Blowfish and IDEA in counter modes
to generate very large "known good" samples. I converted the RAND million
to binary as a 'golden' reference also.

In this game, I was cooking
raw bits (which passed neither Diehard nor MUST) into stuff
that passed both tests ---and was therefore *indistinguishable*
(using these tests) from a known-good PRNG.

During this, I realized that the cipher *designer* plays the *opposite*
game ---the designer tries to get his cipher, given a regular
input (counter mode) or no input (except key and IV, in feedback mode),
to produce this high quality, uniformly distributed bits.

Which is why, again, if you can tell a block-cipher-based-PRNG from a TRNG
without
either the key or taking them apart, let us know.

----------
David Honig

        Inside every wafer is a circuit, yearning to be free.

  


New Message Reply About this list Date view Thread view Subject view Author view

 
All trademarks and copyrights are the property of their respective owners.

Other Directory Sites: SeekWonder | Directory Owners Forum

The following archive was created by hippie-mail 7.98617-22 on Thu May 27 1999 - 23:44:21