A lightweight digital system for the probabilistic characterization of true random bit generators

In this brief paper we present a low-complexity digital architecture for the probabilistic characterization of ergodic TRBGs. Differently from other methods based on the direct estimation of the generation probabilities of finite sets of binary words, in this study we investigate some theoretical relationships between the autocorrelation function of the binary source and the dynamical symbol generation probabilities, that can be exploited to obtain a probabilistic characterization of the TRBG. The solution is proposed to perform a real-time monitoring of TRBGs, that can be suitably implemented in lightweight digital systems.