Re: distinguish TRNG from PRNG

From: AE (hidden@nospam.com)
Date: 03/05/03 

From: AE <hidden@nospam.com>
Date: Wed, 05 Mar 2003 21:31:49 +0100

John E. Hadstate wrote:
> "Carlos Moreno" <moreno_at_mochima_dot_com@xx.xxx> wrote in message
> news:gkc9a.34323$P65.492339@weber.videotron.net...
>
>>Even simpler; forget about radioactive devices: just take
>>thermal noise, low-pass filter it to a manageable frequency
>>range, and measure the times between zero-crossings; if the
>>time between the current (just detected) zero-crossing and
>>the previous one is greater than the time between the previous
>>and the one before the previous, then output 1, otherwise
>>output zero.
>
> Does it have some negative impact on the statistics or the randomness if we
> sample the filtered noise generator at regular intervals and output a 1 or a
> 0 depending on whether the noise signal's slope is positive or not-positive?
>
> Notice that this differs from your proposal in several respects, not the
> least of which it that it outputs a steady stream of (random?) bits.

What if the signat is continuously growing? In that case you'd get a
long sequence of equal bits.

AE

Quantcast