On Fri, 12 Mar 1999 01:27:54 -0500, Sean Breheny <shb7@CORNELL.EDU> wrote:

>At 11:06 AM 3/12/99 +1300, you wrote:
>>Typical processor based RNG's which use an algorithm would be
>>dangerous - they would amenable to cryptographic attack., However, a
>>true random number generator using eg diode thermal noise etc would
>>fit the bill.
>>
>
>Yes,but don't you want something with a uniform distribution? IOW, a one
>time pad where all posible values are equally likely? You can't get this
>directly from diode noise,since it has a non-uniform distribution (I'm
>guessing here,maybe Gaussian or Laplacian?),which would also open it up to
>attack since some numbers would be more likely than others and patterns
>would emerge. You would have to look at the distribution which your diode
>setup produced,and run it through a function which mapped it to a uniform
>distribution (usually in computer algorithms,this is done in reverse,a
>pseudorandom uniform distributed generator is mapped to some other desired
>distribution,such as gaussian).

You'll find no patterns in the thermo noise of a diode. Unless something's wrong
with your circuit, like outputting all ones?

When I've needed truly random patterns in the past, I've used a 16 bit sound
card and sampled the hiss between FM radio stations, and then concatenated the
lowest order bit of the 16 bit words, into a string of bytes.

The lower the sample rate, the better the guarantee the sampled noise will be
truly random.  At too high a sample rate you run into the frequency response of
the radio, at which point consecutive samples may return the same value.

A 1khz sample rate works fine.

-Zonn