A Permutation Challenge Input Interface for Arbiter PUF Variants Against Machine Learning Attacks

Internet of Things (IoT) have broad and deep penetration into our society, and many of them are resource-constrained, calling for lightweight security protocols. Physical unclonable functions (PUFs) leverage physical variations of circuits to produce responses unique for individual devices, and hence are not reproducible even by their manufacturers. Implementable with simplistic circuits and operable with low energy, PUFs are promising candidates as security primitives for resource-constrained IoT devices. Arbiter PUF (APUF) and its variants are lightweight in resource requirements but suffer from vulnerability to machine learning attacks. To defend APUF variants against machine learning attacks, in this paper we investigate a challenge input interface, which incurs low overhead. Analytical and experimental studies were carried out, showing substantial improvement of resistance against machine learning attacks when a PUF is equipped with the interface, rendering interfaced APUF variants promising candidates for security critical applications.