Breaking XOR Arbiter PUFs With Chosen Challenge Attack

Physical Unclonable Functions (PUFs) are lightweight hardware security primitives used for device authentication, identification, and key generation. Among them, XOR Arbiter PUFs were introduced to enhance resistance against modeling attacks by combining multiple Arbiter PUF responses using XOR logic. However, despite this added complexity, XOR Arbiter PUFs remain vulnerable to advanced machine learning and chosen-challenge attacks. This work presents a detailed study on breaking XOR Arbiter PUFs using chosen challenge attacks, highlighting how an attacker can strategically select challenges to efficiently learn internal delay parameters. The study analyzes existing PUF architectures, their limitations, and proposes a secure countermeasure framework to improve robustness against such attacks.