Guidelines for the Design of Random Telegraph Noise-Based True Random Number Generators

Abstract

The development of a robust and secure hardware for the Internet of Things (IoT) and edge computing requires improvements in the existing low-power and low-cost hardware security primitives. Among the various available technologies, true random number generators (TRNGs) that leverage random telegraph noise (RTN) from nanoelectronics devices have emerged as effective solutions. However, the temporal instabilities in the RTN signal, such as the DC drift and temporary inhibition, are a few of the key reliability challenges for the TRNG circuits. In this study, we have utilized experimental RTN data collected from the commonly used gate dielectrics, including silicon dioxide (SiO2), hafnium dioxide (HfO2), and 2D crystalline hexagonal boron nitride (h-BN) to identify the crucial reliability challenges for RTN-based TRNG circuits. We have analyzed the impact of RTN instabilities and of circuit parameters on the output randomness and propose reliability aware design guidelines. Finally, we design and simulate an RTN-based TRNG circuit using a 130 nm CMOS technology and evaluate its reliability at the circuit level.