[Cryptography] The world's most secure TRNG

Bill Cox waywardgeek at gmail.com
Sat Sep 20 22:46:29 EDT 2014


This is a shameless promotion of a dumb I idea I had a while back, the
Infinite Noise Multiplier.  I thought it was cool because it's fast
and simple and easy to integrate on an ASIC, but now that I've become
more of a tin-foil-hat geek, I see that it is in fact the architecture
I prefer for all my true random data.  However... it's only for those
of us who are silly paranoid.

I am building a board-level proof-of-concept, and it uses only a dual
op-amp chip and an dual SPDT switch chip, plus a few caps and
resistors.  It is very simple on a board, and runs way faster than
typical zener noise solutions.

What makes it more secure than other reasonable-speed reasonable-cost
architectures is that it is immune to signal injection.  If you use
zener avalanche noise, for example, or Intel's very cool random
power-up latch state, you are hugely sensitive to signal injection.
These schemes amplify a signal from almost 0V to measurable input
signals.  Any signal-injection attack will dominate over the true
random noise and send the user the attacker's desired data instead.

With an Infinite Noise Multiplier, injected signals simply flip
otherwise random bits of state, which makes them no less random.

So, in case anyone wants to use this, the original idea is at
waywardgee.net.  I hereby renounce any claim to copyrights and patents
related to this project.  I'm giving it away emphatically freely.

I'm working on a board level version over on Upverter.  If there's
interest, I'll post that when it's ready.

Bill
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.metzdowd.com/pipermail/cryptography/attachments/20140920/4ef252cd/attachment.html>


More information about the cryptography mailing list