From: "John A. Thomas" <B858JT@UTARLVM1.UTA.EDU>
Date: Thu, 23 Jun 94 18:30:59 PDT
To: cypherpunks@toad.com
Subject: Hardware RNG's
Message-ID: <9406240130.AA09341@toad.com>
MIME-Version: 1.0
Content-Type: text/plain


Roger, I would like some details about your hardware RNG, if you
could, including how you were reading the random value.  I had the
impression you were taking one bit at a time.

What did you mean when you said:  "I just got a 3 standard deviation
at a 1000000 sample..."?  What statistical testing did you do?

I used the chi-square test on 8-bit values (255 degrees of freedom),
and also computed the probablility for the chi-square statistic.  I
programmed the runs-up and runs-down tests as suggested by Knuth, and
computed the chi-square for those as well.  Finally, I counted the
numbers of 1 and 0 bits, and the number of times 1 followed 0, 0
followed 1, etc.

I didn't do the autocorrelation test.  If you have any code for that,
I would appreciate having it.  I'm not sure what you mean by "...the
derevitives of the sequence..."  What is that?

I suppose these gadgets could be useful for those who want the
absolute security of the one-time pad, and can exchange disks
securely.  With disks holding 1.44 meg now, it could be practical.

John A. Thomas
b858jt@utarlvm1.uta.edu
75236.3536@compuserve.com
PGP public key available.