Physical unclonable function (PUF) is considered as a promising primitive for variety of security applications since its appearance, such as authentication, key generators and random oracle. With the fast development of Internet of Things (IoT), the requirement for low complexity, high power efficiency and enhanced security level of PUF becomes urgent. Meanwhile, the conventional PUF instances based on optical effect and semiconductors fail to meet the requirements due to poor scalability and sensitivity to attacks. This paper proposes a circuit design of PUF by utilizing the spin transfer torque magnetic random access memory (STT-MRAM). The performance of presented circuit design is thoroughly investigated from the aspects of uniqueness, reliability, uniformity, diffuseness, access speed and energy consumption. The simulation results show that the inter-Hamming distance can achieve 50% and the intra-Hamming distance can be reduced as low as 0%, resulting in perfect security. Meanwhile, the circuit behaves high immunity to environmental variations such as high and low temperature. With the rapid development of artificial intelligence, many powerful attack methods emerge. To defend against these attacks, the user reconfigurable function has been proposed. Moreover, design space including programming voltage, device parameters and thermal conditions has been explored to optimize the reliability of this function.

中文翻译：

---

基于STT-MRAM的物理不可克隆函数的可靠性分析与性能评估

物理不可克隆函数 (PUF) 自出现以来被认为是各种安全应用的有前途的原语，例如身份验证、密钥生成器和随机预言机。随着物联网(IoT)的快速发展，对PUF低复杂度、高能效和增强安全级别的要求变得迫切。同时，传统的基于光学效应和半导体的PUF实例由于扩展性差和对攻击的敏感性差，无法满足要求。本文提出了一种利用自旋转移矩磁性随机存取存储器（STT-MRAM）的PUF电路设计。从唯一性、可靠性、均匀性、扩散性、访问速度和能耗等方面对所提出的电路设计的性能进行了深入研究。仿真结果表明，Hamming 间距离可以达到 50%，Hamming 内距离可以降低到 0%，从而实现了完美的安全性。同时，该电路对高温和低温等环境变化具有很高的免疫力。随着人工智能的飞速发展，出现了许多强大的攻击手段。为了防御这些攻击，已经提出了用户可重构功能。此外，还探索了包括编程电压、器件参数和热条件在内的设计空间，以优化该功能的可靠性。随着人工智能的飞速发展，出现了许多强大的攻击手段。为了防御这些攻击，已经提出了用户可重构功能。此外，还探索了包括编程电压、器件参数和热条件在内的设计空间，以优化该功能的可靠性。随着人工智能的飞速发展，出现了许多强大的攻击手段。为了防御这些攻击，已经提出了用户可重构功能。此外，还探索了包括编程电压、器件参数和热条件在内的设计空间，以优化该功能的可靠性。