image: (lower right) reconfigurability enhances security and preserves the privacy of the previous owner when tokens are exchanged; (left) proposed PUF demonstrates high reliability across different environments and resists machine-learning attacks. BL: bit line; SL: source line.
Credit: Xiuye Zhang et al.
In a recent development published in Engineering, researchers from Beihang University and Truth Memory Corporation have fabricated a 1 Kbit spin-orbit torque magnetic random access memory (SOT-MRAM) chip using a 180 nm complementary metal oxide semiconductor (CMOS) process, and implemented a physical unclonable function (PUF) on it, presenting a potential solution for Internet of Things (IoT) security challenges.
The exponential growth of the IoT has brought about significant hardware security issues. Constrained by cost, IoT devices often lack robust encryption mechanisms, making private data vulnerable to attacks. PUFs, which can serve as the root of trust for IoT devices, have emerged as a promising lightweight cryptographic solution. However, conventional CMOS-based PUFs have drawbacks such as insufficient randomness, high power and area consumption, and vulnerability to environmental factors.
The newly developed SOT-MRAM sr-PUF addresses these problems. It achieves a strong, highly reliable, and reconfigurable PUF that can resist machine-learning attacks. The PUF is initialized by setting a specific writing voltage to balance the probabilities of high- and low-resistance states in the memory array to around 50%. Then, a computing-in-memory (CIM) approach is used to generate responses. By comparing the current summations according to different column combinations, a 1-bit response is generated.
The performance of the SOT-MRAM sr-PUF is remarkable. It has a challenge-response pair (CRP) capacity of 109. Its uniformity is 50.07%, diffuseness is 50%, uniqueness is 49.89%, and the bit error rate is 0%, even in a 375 K environment. These values are near-ideal, indicating excellent randomness and reliability.
Moreover, the reconfigurability of the PUF is a key feature. By applying different write voltages, the PUF can dynamically refresh its CRPs. The reconfigurable Hamming distance is 49.31%, and the correlation coefficient between different reconfiguration cycles is less than 0.2. This makes it difficult for attackers to extract output keys through side-channel analysis.
When it comes to machine-learning attacks, the SOT-MRAM sr-PUF shows great resilience. Tested with three machine-learning algorithms (logistic regression, support vector machine, and multilayer perceptron), its prediction accuracy is close to the ideal 50%, meaning it can effectively resist such attacks.
This research represents a significant step forward in the field of PUFs. The SOT-MRAM-based PUF offers a practical and reliable hardware security solution for future IoT applications, potentially enhancing the security of countless IoT devices.
The paper “Experimental Realization of Physical Unclonable Function Chip Utilizing Spintronic Memories,” authored by Xiuye Zhang, Chuanpeng Jiang, Jialiang Yin, Daoqian Zhu, Shiqi Wang, Sai Li, Zhongxiang Zhang, Ao Du, Wenlong Cai, Hongxi Liu, Kewen Shi, Kaihua Cao, Zhaohao Wang, Weisheng Zhao. Full text of the open access paper: https://doi.org/10.1016/j.eng.2024.12.028. For more information about the Engineering, follow us on X (https://twitter.com/EngineeringJrnl) & like us on Facebook (https://www.facebook.com/EngineeringJrnl).
Journal
Engineering
Article Title
Experimental Realization of Physical Unclonable Function Chip Utilizing Spintronic Memories
Article Publication Date
3-Jan-2025