Other Directory Sites: SeekWonder | Directory Owners Forum
The following archive was created by hippie-mail 7.98617-22 on Fri Aug 21 1998 - 17:20:09 ADT
Enzo Michelangeli (em@who.net)
Tue, 7 Jul 1998 19:00:15 +0800
-----Original Message-----
From: Ben Laurie <ben@algroup.co.uk>
To: Antonomasia <ant@notatla.demon.co.uk>
Cc: CodherPlunks@toad.com <CodherPlunks@toad.com>
Date: Tuesday, July 07, 1998 6:24 PM
Subject: Re: Random Data from Geiger Counter
[...]
>I was suggesting using the length of the intervals to generate more
>bits, rather than the count in a specified time.
In fact, I can see no reason why the number of bits cannot be made as large
as allowed by the resolution of the time counter: after all, the time of
decay is not quantized. For example, if your digital counter is clocked at
100 MHz and gated by the decay events, about one decay per second yields
counts about 26 bit long. Discard, say, the most significant 10 bits, and
you'll get something very close to 16 bits of entropy per decay event. The
distribution should be extremely close to uniform, because in a Poisson
statistics the chances of getting periods 2^10 times smaller than the mean
value (which would skew the distribution) are negligible. To make it safer,
discard them after performing a hash.
My point is that the constraints here seem to be only practical (frequencies
too high for the time counter to accept, or periods too long for a
comfortable measurement), not theoretical.
Enzo
The following archive was created by hippie-mail 7.98617-22 on Fri Aug 21 1998 - 17:20:09 ADT