Subject: Re: Hardware random number generator
From: ted@nmsu.edu (Ted Dunning @ NMSU Computer Science)

In article <13832@ulysses.att.com> smb@ulysses.att.com (Steven Bellovin) writes:
   Path: nmsu!ariel.unm.edu!pprg.unm.edu!topgun!lll-winken!decwrl!ucbvax!ulysses!ulysses.att.com!smb
   From: smb@ulysses.att.com (Steven Bellovin)
   Newsgroups: sci.crypt
   Date: 4 Oct 90 01:50:43 GMT
   References: <1990Oct3.205653.150@cs.utk.edu>
   Sender: netnews@ulysses.att.com
   Lines: 11

   There are a few things to be careful of, such as the ratio between the
   oscillator frequencies -- if they're too close, you don't get random
   enough bits.  Also, be *very* careful to avoid coupling between the
   two oscillators.


one interesting approach to avoiding this problem would be to generate
pink noise and inject it into the timing chain on the low frequency
oscillator.  with would have the effect of randomizing the timings
involved and would tend to break any coherence effects.

--
ted@nmsu.edu      +---------+
                  | In this |
                  |  style  |
                  |__10/6___|