A 0.186-pJ per Bit Latch-Based True Random Number Generator with Mismatch Compensation and Random Noise Enhancement
Ruilin Zhang, Xingyu Wang, Luying Wang, Xinpeng Chen, Fan Yang, Kunyang Liu, Hirofumi Shinohara · 2021
A calibration and feedback control-free latch-based true random-number generator (TRNG) is presented. It features a mismatch self-compensation and a random noise enhancement technique to drastically improve the noise-to-mismatch ratio. By employing the XOR function of only 4-bit entropy sources, the proposed TRNG can efficiently operate across a wide voltage (0.3~1.0 V) and temperature (−20~100°C) range. An 8-bit von Neumann with waiting (VN8W) post-processing technique is used to extract full entropy bitstreams, which have been verified by the NIST-SP 800-22 randomness tests. Robustness against supply noise injection attack is also demonstrated. The proposed TRNG is fabricated in 130-nm CMOS technology and achieves the state-of-the-art energy of 0.186 pJ/bit at 0.3 V with a core area of 661 um2(0.039 MF2).