Self-timed Ring based True Random Number Generator: Threat model and countermeasures

Abstract

Self-timed Ring based True Random Generators (STRNGs) extract randomness from the jitter of events evenly propagating in a Self-Timed Ring (STR) oscillator. Security of such generators is primarily based on an entropy assessment: an accurate model of the minimum entropy per output bit with physical measurement of the noise source. This assessment is reinforced with both entropy source monitoring and online testing of the output bits. This paper addresses the security of the STRNG. First we identify potential vulnerabilities on the generator and define a threat model. Based on this threat model, we analyze the effect of active attacks in analog simulations (in a 55 nm technology), and by emulating them in a high-level simulation model. Then, we propose simple and efficient countermeasures to thwart attacks focusing on the generator. Finally, we evaluate the output sequences before and after attacks to validate the proposed countermeasures.