Efficient PUF Error Correction through Response Weighting

Yuejiang Wen, Yingjie Lao

Abstract: Reliability is an important performance metric of physical unclonable function (PUF) based authentication. This paper proposes a novel methodology that incorporates error rates into PUF response error correction to improve the overall performance. The proposed method first obtains the PUF circuit parameters by using machine learning techniques, which are then used to estimate the corresponding error rates. Then, we assign PUF response bits with different degrees of error-tolerance, according to their estimated error rates. Response weighting algorithm is proposed to determine the optimal weight assignment for each PUF response bit, which is formulated as an integer optimization problem. Experimental results show that the proposed response weighting algorithm can reduce not only the false negative from 20.6% to 8.3% under a noisy environment, but also the false positive rate by 58% for a PUF-based authentication with 127-bit response and 13-bit error correction.