Hybrid Machine Learning Attack for Feed-Forward Arbiter PUF and Its Evaluation

Abstract

The physical unclonable functions (PUFs) have attracted attention to ensure the security of internet of things (IoT) devices. On the other hand, the threat of machine learning attacks is pointed out; therefore, the feed-forward arbiter (FFA) PUF has been proposed as the resistant PUF. This study proposes a new machine learning attack for the FFA PUF. The proposed method focuses on both power consumption generated during the operation and the selectable challenge, and a hybrid machine learning attack which combines them are introduced to predict the response of the FFA PUF. Experiments using a field programmable gate array evaluate the validity of the proposed method.