True random number generator combined with SAR ADC

Abstract

Random number generators are essential hardware for generating cryptographic keys for information security. Among them, true random number generators, which generates random numbers using an entropy source, have the advantage of unpredictability compared to pseudo random number generators, which generate random numbers by algorithms, but have the disadvantage of power efficiency. This paper points out several problems of the existing research of true random number generators, and proposes an improved circuit method. Based on the method of random number generator for a research project that utilized the output voltage of silicon nanowire devices and the quantization noise of analog-to-digital converters as entropy sources, the proposed architecture is random number generator architecture using two-step successive approximate register analog-to-digital converter(SAR ADC). By repeating the process of returning the coarse quantization noise of the analog-to-digital converter to the next input, randomness is increased. It also has the advantage that signal conversion and random number generation can be performed sequentially in the same converter, and the performance of each operation is independent.