A 10-Gb/s True Random Number Generator Using ML-Resistant Middle Square Method
Jonghyun Kim, Hyungil Chae
Abstract
A 10-Gb/s throughput all-digital true-random-number generator (TRNG) is introduced. The TRNG comprises an entropy source based on a StrongARM latch comparator and a post-processing block using the middle square method (MSM). The comparator is fully digital, so the entropy source can operate at a high frequency with robustness to process-voltage-temperature (PVT) variation. The MSM post-processing effectively removes any colored noise from the entropy source. The randomness of the MSM post-processing is evaluated by machine-learning resistance. The prototype TRNG is fabricated in a 28-nm CMOS process, and the measurement shows a 10-Gb/s throughput with a power efficiency of 0.121 pJ/bit. The output random bit stream passed all the NIST SP 800-22 and NIST SP 800-90B tests under various operating conditions with a maximum Shannon entropy of 0.999999999997.