Standard or guidance for entropy collection

Question

I need to write some high-level security requirements which include a random number generator suitable for use in cryptography. I don't need something very formal, this isn't a security certification or a work order. I just want to go to the next level from “supply a proper RNG”.

A proper RNG should be built on a cryptographic PRNG seeded from an appropriate entropy source. For the cryptographic PRNG and even the seeding method, it isn't hard to find applicable standards; for example, I'd be content to write that the PRNG must use one of the algorithms listed in FIPS 140-2 annex C, or attain an equivalent security level.

What about entropy collection or seeding? Is there a similar document that I can quote? Anything easily available in English would do, it doesn't have to be (or not to be) a government standard or an RFC or similar, but it's better if it looks somewhat official or trustworthy.

Offering concrete guidance would be a plus. However this could be hard since the requirements may need to be met in very different ways (e.g. a non-deterministic oscillator in an embedded device, a server with a radioactive source, a PC with someone wiggling the mouse, …).

Answer 1

As far as I can tell, NIST has only one official document about entropy collection. SP-800-90B.

The purpose of NIST Special Publication (SP) 800-90B is to specify the design and testing requirements for entropy sources that can be validated as approved entropy sources by NIST‘s CAVP and CMVP.

It essentially defines a bunch of statistical tests to check the strength of a particular entropy source. Seeking entropy methods that meets the requirements is a good first step especially if you need "official" certification by NIST.

Answer 2

Germany's BSI has produced AIS 31 that includes requirements on Physical True RNGs (PTRNGs). It is designed to fill a gap in the Common Criteria standard.

Chapter 4 describes pre-defined classes for physical true, non-physical true, deterministic and hybrid random number generators. ... The basic concepts and evaluation criteria are illustrated by additional examples in chapter 5.