2
$\begingroup$

NIST SP 800-22 (published in 2010) lists out some statistical tests for telling good random bit generators from bad ones. NIST SP 800-90B (published in 2018) has apparently the same purpose but lists out a very different set of tests. Why?

$\endgroup$
3
$\begingroup$

Publication NIST SP 800-22 is a Standard, the NIST SP 800-90B recommendation might someday replace the current standard after it is ratified.


Sources for answer:

NIST SP 800-22: "A Statistical Test Suite for Random and Pseudorandom Number Generators for Cryptographic Applications", Published: September 16, 2010

Abstract
This paper discusses some aspects of selecting and testing random and pseudorandom number generators. The outputs of such generators may be used in many cryptographic applications, such as the generation of key material. Generators suitable for use in cryptographic applications may need to meet stronger requirements than for other applications. ...
[Supersedes SP 800-22 (May 15, 2001): http://www.nist.gov/manuscript-publication-search.cfm?pub_id=151222 ].

News Release: "NIST is Proud to Announce the Release of 2 DRAFT Publications" (September 05, 2012):

NIST requests comments on two Draft publications for random bit generation: Draft SP 800-90B, Recommendation for the Entropy Sources Used for Random Bit Generation and Draft SP 800-90C, Recommendation for Random Bit Generator (RBG) Constructions.

News Release: "NIST Requests Comments on Computer Security Publication on Randomness" (January 27, 2016):

The Second Draft of Special Publication (SP) 800-90B, Recommendation for the Entropy Sources Used for Random Bit Generation, aims to help security specialists judge whether the source of random numbers they use as part of the data encryption process is sufficiently unpredictable. NIST is requesting public comments by May 9, 2016, on the draft document, which is available at NIST's CSRC website.

Series 800-90 Documents: "Random Bit Generation":

Project Overview

The following publications specify the design and implementation of random bit generators (RBGs), in two classes: Deterministic Random Bit Generators (pseudo RBGs); and Non-Deterministic Random bit Generators (True RBGs).

  • SP 800-90A, Recommendation for Random Number Generation Using Deterministic Random Bit Generators

June 25, 2015: This Recommendation specifies mechanisms for the generation of random bits using deterministic methods. In this revision, the specification of the Dual_EC_DRBG has been removed. The remaining DRBGs (i.e., Hash_DRBG, HMAC_DRBG and CTR_DRBG) are recommended for use. Other changes included in this revision are listed in an appendix.

  • SP 800-90B, Recommendation for the Entropy Sources Used for Random Bit Generation

January 10, 2018: This Recommendation specifies the design principles and requirements for the entropy sources used by Random Bit Generators, and the tests for the validation of entropy sources. These entropy sources are intended to be combined with Deterministic Random Bit Generator mechanisms that are specified in SP 800-90A to construct Random Bit Generators, as specified in SP 800-90C.

  • SP 800-90C, Recommendation for Random Bit Generator (RBG) Constructions

April 13, 2016: NIST invites comments on the second draft of NIST Special Publication (SP) 800-90C, Recommendation for Random Bit Generator (RBG) Constructions. This Recommendation specifies constructions for the implementation of RBGs. An RBG may be a deterministic random bit generator (DRBG) or a non-deterministic random bit generator (NRBG). The constructed RBGs consist of DRBG mechanisms, as specified in SP 800-90A, and entropy sources, as specified in SP 800-90B. The comment period closed June 13, 2016

On May 2-3, 2016, NIST hosted a workshop on Random Number Generation to discuss the SP 800-90 series of documents--specifically, SP 800-90B and SP 800-90C.

| cite | improve this answer | |
$\endgroup$
  • 1
    $\begingroup$ The question is on the statistics tests themselves contained in these two documents. For example, from 2010 to 2018 has there been a revolution in the way RBGs are tested? Why have the tests changed so much from 2010 to 2018? (For instance, in NIST 800-22 there's the Binary Matrix Rank Test and the Discrete Fourier Transform Test, neither of which are present in NIST 800-90B. Why not?) $\endgroup$ – user45491 Sep 12 '18 at 12:26
  • $\begingroup$ Many numerical sequences will pass the BMRT, including files that have been compressed and the theoretical reference distribution of the DFTT statistic has not been derived. Better tests than the DFTT have been developed: Randomness Test to Solve Discrete Fourier Transform Test Problems. Despite dissatisfaction with 800-22 it still must be passed to obtain that certification. Whether the new list is sufficient is under discussion: nist.gov/news-events/events/2016/05/… $\endgroup$ – Rob Sep 12 '18 at 16:17

Not the answer you're looking for? Browse other questions tagged or ask your own question.