Hi Steffen,
> i need to create some strong random numbers
This is always a considerable challenge, no?
> Possibly someone of you got any idea what sources of noise i could take.
For the time being, the most reasonable source (for me) has been interrupts that
are asynchronous/aperiodic in nature (e.g. uart, ethernet, etc). But that's just
half (or less) of the problem: how you use them can be a headache as well ;-)
Combining an async driven LFSR with a clock-based LFSR as a salt source for
a crypto algorithm (the output of which is the random data) appears good -- but
I never did the math 8-P
Please don't be shy about posting your ideas -- this is very, very challenging stuff!
Regards,
--Scott