[ale] OT: Quantifying Randomness

[Glenn C. Lasher Jr.]

You could try a frequency analysis.  Take your data and push them into a
series of 8-bit values. Collect 256*n values.  Use a number of different
values for n, as this will cause different patterns to emerge, if there
are any.  The higher the value of n, the longer-term the pattern that
emerges will be.

Next, you count the number of instances for every possible value between 0
and 255 inclusive.  Graph this as frequency vs value.  If you have a good,
even distribution, then you will have a more-or-less horizontal line.

Now, it is important to note, however, that this will not give you a
guarantee that there are no patterns.  A counter that ran from 0-255 would
give a perfect result.  This will, however, help ensure that you have a
fairly even distribution of results.


On 4 Jul 2002, Kevin Krumwiede wrote:

> I'm looking for a way to ensure that random data captured from
> recordings of white noise does not contain patterns induced by the
> equipment.  I've seen programs that analyze the output of a PRNG and
> quantify its randomness, but I don't know what they're called and I
> can't find anything on Google.  If someone can point me to a white paper
> or an algorithm, I can implement it...
>
> Thanks,
> Krum
>
>
> ---
> This message has been sent through the ALE general discussion list.
> See http://www.ale.org/mailing-lists.shtml for more info. Problems should be
> sent to listmaster at ale dot org.
>

glasher at nycap.rr.com
You've been programmed by the Illuminati not to see the word "".



---
This message has been sent through the ALE general discussion list.
See http://www.ale.org/mailing-lists.shtml for more info. Problems should be 
sent to listmaster at ale dot org.