[Cryptech Tech] User auditable hardware entropy source/random number generator

[Benedikt Stockebrand]

Hi Fredrik and list,

first off: Enjoy your holiday while you can---if you eventually blame me
for being unable to relax and recover, I'll refuse to take any
responsibility:-) Just let me know when you're around again and I'll
give you an up-to-date schematic and whatever documentation I have by
then.

Fredrik Thulin <fredrik at thulin.net> writes:

> I am more than interested. 

Cool.  I've just spent an extended lunch break messing around with some
new types of Zener diodes I got delivered this morning; looks like I've
been exceedingly---and I mean exceedingly---lucky with the ones I used
for the tests so far.

I know that the NSA was good at pulling all sorts of strings, but the
way they managed to sneak in all these ultra-low-noise Zener diodes is
really impressive on the border of scary:-)

Anyway, aside from the ones I originally used (from a set, without
proper type information beyond the labelling on the diodes themselves) I
have three types of Zener diodes that seem to work.  Unfortunately they
are all SMD only. 

Eventually I'll have to make up my mind if I rather want to stick with a
limited choice of Zener diode types or add another transistor stage
again to make the circuit work with just about any make of Zener diode.
Hmm.

> I too was planning to use a microcontroller to sample the avalanche noise and 
> then send the entropy on to the FPGA to use as one source in Joachims mixer. I 
> do expect some debate around the trust issues with sticking another 
> microcontroller in there, but that would have been my starting point at least.

Now, the reason why I use a microcontroller here is because I am kind of
familiar with them, while I've never actually worked with FPGAs.  If
anybody can do the same job within an FPGA, that's all the better.

It's just that I can't really tell if this is feasible, and by starting
this project I've already opened quite a few cans of worms^W^W^W
opportunities to expand my horizon, so I deferred FPGAs to whenever I
got this project up and running (seriously).

Anyway, in my design the MCU here has two jobs: To massage the output
from the entropy source/noise generator circuit into a proper random
bitstream (no bias, no correlation between bits) and to pass it on
towards the computer.  The first job must be done somewhere, but the
second one may become irrelevant if the FPGA is receiving the output.


Cheers,

    Benedikt

-- 
Benedikt Stockebrand,                   Stepladder IT Training+Consulting
Dipl.-Inform.                           http://www.stepladder-it.com/

          Business Grade IPv6 --- Consulting, Training, Projects

BIVBlog---Benedikt's IT Video Blog: http://www.stepladder-it.com/bivblog/