[Gllug] entropykey: why did nobody ever mention this thing before?

[Richard Jones]

On Mon, Aug 02, 2010 at 10:32:19AM +0100, Richard Jones wrote:
> The clock rate that you have to sample at must be some large
> multiple of this frequency in order to reduce correlation with the
> frequency of the diode

I didn't explain that very well.

If you have a source which switches on and off at random, you can
measure the average frequency - say 100 Hz.  If you sample this at
some rate, then there is a probability that adjacent bits are
correlated.  eg. If you sampled this source at 10,000Hz then often
adjacent samples would be the same.  You have to sample at a much
lower frequency (eg. 1 Hz) to reduce the probability that bits are
correlated.

You can work out mathematically the exact probability.  And no matter
how slow you sample, the probability is always > 0.  You can reduce it
to below any acceptable probability, but your sample rate, hence the
number of random bits you get out, hence the usefulness of your
device, becomes very low.

Rich.

-- 
Richard Jones
Red Hat
-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug