Bias-Independent True Random Number Generator Circuit using Memristor Noise Signals as Entropy Source

Abstract

Inherent noise characteristics of memristor devices can be utilized in stochastic computing applications such as true random number generators (TRNGs). However, the ratio between capture and emission time can significantly affect the randomness of generated bit streams by TRNGs. Herein, a bias-independent TRNG circuit is presented, utilizing the random telegraph noise (RTN) signal of the memristor as a random entropy source. This design considers the condition-dependent RTN characteristics, including capture time and emission time constants, which vary with read voltage (Vread) and temperature conditions in the high-resistance state of the fabricated memristor. The TRNG circuit, comprising an edge detection circuit and an N-bit counter, is experimentally demonstrated to validate hardware feasibility with the optimized external clock frequency, which can mitigate the biases induced by Vread and temperature. Finally, the performance of the designed TRNG circuit is evaluated using autocorrelation functions and National Institute of Standards and Technology tests, confirming its capability to produce random number bitstreams.