How Random is Your RNG?

Published

January 18, 2015

Author(s)

Meltem Sonmez Turan, John M. Kelsey, Kerry A. McKay

Abstract

Cryptographic primitives need random numbers to protect your data. Random numbers are used for generating secret keys, nonces, random paddings, initialization vectors, salts, etc. Deterministic pseudorandom number generators are useful, but they still need truly random seeds generated by entropy sources in order to produce random numbers. Researchers have shown examples of deployed systems that did not have enough randomness in their entropy sources, and as a result, crypto keys were compromised. So how do you know how much entropy is in your entropy source? Estimating entropy is a difficult (if not impossible) problem, and we've been working to create usable guidance that will give conservative estimates on the amount of entropy in an entropy source. We want to share some of the challenges and proposed methods. We will also talk about some new directions that we're investigating, and present results of our estimation methods on simulated entropy sources.

Proceedings Title

Shmoocon Proceedings

Conference Dates

January 16-18, 2015

Conference Location

Washington, DC

Conference Title

Shmoocon

Pub Type

Conferences

Download Paper

Local Download

Keywords

entropy, random number generation, RNG, prediction

Citation

Sonmez, M. , Kelsey, J. and McKay, K. (2015), How Random is Your RNG?, Shmoocon Proceedings, Washington, DC, [online], https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=917957 (Accessed October 10, 2025)