Reconfigurable Physical Unclonable Functions Based on Magnetic Domain Patterns

Abstract

Abstract Ensuring authentication, data security, and privacy has become critically important with the rise of Internet of Things (IoT) systems. Physical unclonable functions (PUFs) employ the intrinsic physical randomness of materials to generate unique, nonreplicable outputs, ensuring device authenticity and protection against cloning. Although silicon‐based PUFs have significantly advanced, developing practical and cost‐effective solutions for resource‐constrained embedded systems remains challenging. This study demonstrates a reconfigurable PUF design that leverages spontaneously formed labyrinth domain patterns in magnetic thin films as a high‐entropy source, without requiring any external stimuli or control. The resulting PUFs display nearly ideal performance in uniqueness, uniformity, bit‐aliasing, and robustness, underscoring their practicality and reliability. These PUFs demonstrate high indivisibility in linear feature spaces and robustness against compromising the security or functionality of machine learning. More importantly, the electrical readout of domain‐based PUF signals is demonstrated using magnetic tunnel junctions and propose an architecture for integrating the PUF system with existing silicon‐based circuits. Overall, the proposed PUF system offers a practical and innovative solution for advanced anti‐counterfeiting measures and secure device identification.