Extracting more entropy for TRNGs based on coherent sampling

Abstract

True Random Number Generators (TRNGs) are essential for cryptographic systems and communication security. According to the published standards, sufficient entropy derived from the stochastic model is required for TRNGs. Compared with the directly sampling jittery oscillating signal, the coherent sampling is a more efficient entropy extraction technique. In this paper, under the premise that the entropy per bit is sufficient, we focus on how to extract the entropy as much as possible from the coherent sampling in order to enhance the throughput of TRNGs. We provide a parameter adjustment method to maximize the generated entropy rate, and this method is based on our proposed stochastic model. According to the method, we design a TRNG architecture and implement it in Field Programmable Gate Arrays (FPGAs). In the experiment, the improved generation speed is up to 4 Mbps, and the output sequence is able to pass NIST SP 800-22 statistical tests without postprocessing. Compared to the basic coherent sampling, the bit generation rate is improved to 12 times.