[Noisebridge-discuss] Group order of Entropy Keys: hardware random number generators

[Jonathan Foote]

Well, like I said, it's a noise source into a whitening algorithm, see
the nice diagram Miloh found.

I guessed wrong in that the noise source is junction noise rather than
thermal noise (a little surprising as I recall thermal noise is flat
down to DC while junction noise is frequency-dependent). Also an
amplified junction is an AM radio, I guess the "isolation/shielding"
helps with the EMI, but I could certainly see that as a possible
attack.

Not going to say we could build anything truly random, but looks like
Maurer's algorithm can estimate the entropy which is a nice sanity
check.

Putting the noise back in noisebridge, how about a web service
delivering quantum entropy to the world? Useful for one-time pads,
blinkenlichts für looksenpeepers, or robot worship a la Rudy Rucker.

On Tue, Jun 19, 2012 at 10:32 PM, Brian Cloutier
<briancloutier2010 at gmail.com> wrote:
>
>> I'm guessing the entropykey is nothing more than a little amp for
>> thermal (resistor) noise into an ADC, followed by a whitening
>> algorithm. And some kind of device driver to pipe it into /dev/random.
>
> Its quite a bit more than that. According to the website its a transistor
> with some reverse voltage applied /just/ to the the point where it breaks
> down. Electrons that escape through "quantum tunneling" (I don't know enough
> ee to know of this is BS or not, but it sounds like the correct effect) are
> then measured. Its actually two transistors, each of which are measured for
> entropy, which are combined into a random bit source which is again measured
> for entropy. Apparently the stream is then encrypted, the USB key emulates a
> serial port, and a small program on the host decrypts the data and finally
> feeds it into the entropy pool. (I'm at a loss as to why its encrypted, but
> I assume its to prevent some obscure side channel attack like reading the
> USB line voltage off an LED on the case flickering and inferring the data
> sent)
>
>> Of course we could make a radioactive one with Mike Kan's transistor
>> alpha detector trick (pop the top off an old TO-3 can transistor),
>
> I would absolutely /love/ to do this, especially if we hook it into a banana
> (potassium is radioactive!) as our source of radiation. We could then make
> jokes along the lines of "well how are /you/ going to feed a million
> simulated monkeys bashing randomly? A mere /pseudo/random source?"
>
> - Brian
>
>
> _______________________________________________
> Noisebridge-discuss mailing list
> Noisebridge-discuss at lists.noisebridge.net
> https://www.noisebridge.net/mailman/listinfo/noisebridge-discuss
>