[Cryptography] The Crypto Pi

[Arnold Reinhold]

> On Jan 16, 2015, at 10:16 AM, Henry Baker <hbaker1 at pipeline.com> wrote:
> 
> At 04:29 PM 1/15/2015, Arnold Reinhold wrote:
>> A camera might be cheaper than a plug-in HRNG, doesn't take up a USB port, and could be easier to audit.
> 
> A fixed security camera will power up always seeing the same scene, so some non-trivial effort is required to ensure randomness.
> 

There are two potential sources of randomness in a digital camera image, randomness in the scene being imaged and randomness in the picture taking process, e.g. from quantization noise, thermal noise, etc. We discussed last year various ways to insure the former (M&Ms or other small objects on a vibrating plate, a bubble wall in a water tank, tropical fish, Sun’s patented Lava Lamp, ...) The later needs some experimental testing. How close to identical are two successive image captures of ostensibly the same scene, or alternatively taken with the lens cap on? How much entropy is in the difference? No point in speculating, the Raspberry Pi seems an ideal platform to do such tests.

Arnold Reinhold