Leakage-resilient memory-based physical unclonable function using phase change material

Abstract

Memory-based Physical Unclonable Function (MemPUF) emerged as a replacement for traditional key preservation primitives to overcome the susceptibility of secret keys to physical attacks. Recent experiments demonstrated that even some MemPUFs can be physically attacked by exploiting their side-channel information. In this paper, we formulate an adversary model for a prediction attack that takes advantage of the side-channel information leaked from a MemPUF. Based on this pivotal insight, we propose countermeasures to enhance the resilience of MemPUFs against such a kind of attack, and introduce a security-enhanced MemPUF design using phase change material. Our analysis demonstrated the effectiveness of our proposed scheme against the measurement-prediction attack given an adversary with certain bounded attack capability.