[117600] in Cypherpunks
Re: otp
daemon@ATHENA.MIT.EDU (John A. Limpert)
Mon Sep 6 20:58:33 1999
Message-ID: <37D45C78.FC76E780@radix.net>
Date: Mon, 06 Sep 1999 20:29:44 -0400
From: "John A. Limpert" <johnl@radix.net>
MIME-Version: 1.0
To: cypherpunks@cyberpass.net
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Reply-To: "John A. Limpert" <johnl@radix.net>
Anonymous wrote:
>
> At 10:24 AM 9/6/99 -0700, Tim May wrote:
> >The dead time of the Geiger tube will stop this from working. Geiger
> >counters are a poor choice for any such uses.
I am aware of the dead time problem. The Geiger counter that I am using
has an additional problem. The HV power supply appears to collapse at
very high count rates.
> Clearly, radioactive sources are for hobbyists only.
> And although decay is "perfectly random" no measurement
> can preserve this. One is better off with an audio source.
> Or if one likes that kind of tinkering, build a circuit: check the patents
> for this.
My software measures the time interval between decays. A random bit is
generated from comparing the values of two successive measurements. The
result passes statistical tests for entropy, bias and correlation. Most
of this is based on the hotbits random number generator.
I've seen circuits based on Johnson noise and avalanche noise. One of my
teachers (an EE) told me that it is difficult to build a noise source of
cryptographic quality. There are many ways of screwing it up. He said
that the circuit had to be carefully laid out, with good shielding and
power supply isolation. Even then, getting rid of bias was difficult.
--
John A. Limpert johnl@radix.net PGP keyID 0x15D681D0
