The generation of true random numbers is one of the most important tasks in a hardware security module (HSM), particularly for cryptography applications. The stochastic behavior of electronic devices can be used to generate random numbers. In this paper, a reliable neuromorphic true random number generator (TRNG) relying on stochastic switching of the magnetic tunnel junction (MTJ) in the subcritical current regime is proposed. Thanks to the efficient structure of the proposed design as well as the fascinating features of the MTJs and carbon nanotube field-effect transistors (CNTFET), the proposed TRNG consumes low power. Moreover, the neuromorphic structure of the proposed design leads to variation tolerance and guarantees the truly random number generation even in the presence of the process variations. HSPICE simulations verify the functionality of the proposed TRNG. Furthermore, by considering the corners of the fabrication process, the randomness of the bitstream, generated by the proposed TRNG, is validated by the statistical randomness test provided by the U.S National Institute of Standards and Technology (NIST).

中文翻译：

---

基于神经形态变异容忍自旋电子结构的可靠硬件安全模块的真随机数发生器

真随机数的生成是硬件安全模块 (HSM) 中最重要的任务之一，特别是对于密码学应用。电子设备的随机行为可用于生成随机数。在本文中，提出了一种可靠的神经形态真随机数发生器（TRNG），它依赖于亚临界电流状态下磁隧道结（MTJ）的随机切换。由于所提议设计的高效结构以及 MTJ 和碳纳米管场效应晶体管 (CNTFET) 的迷人特征，所提议的 TRNG 功耗低。此外，所提出的设计的神经形态结构导致了变化容限，即使在存在过程变化的情况下也能保证真正的随机数生成。HSPICE 仿真验证了提议的 TRNG 的功能。此外，通过考虑制造过程的角落，由美国国家标准与技术研究所 (NIST) 提供的统计随机性测试验证了由提议的 TRNG 生成的比特流的随机性。