PUF Authentication Scheme Enhanced by Multi-path Fading Channel Attributes

With the booming growth of the Internet of Things (IoT), resource-constrained devices play an essential role in our daily lives. Ensuring the secure access of resource-constrained devices is of paramount importance. Physical unclonable function (PUF) authentication has been proposed as a promising solution for preventing unauthorized access of resource-constrained devices due to its low complexity. However, the mapping relationship between the challenge and response pairs (CRPs), which is fundamental to the PUF authentication, can be easily predicted by modeling attacks based on machine learning (MAML). An adversary can use the predicted CRPs to perform spoofing attacks, severely deteriorating the efficiency of networks. To solve the problem, we propose a channel and PUF fusion authentication (CPFA) scheme to defend against MAML. Specifically, the multi-path fading attributes (MFA) are introduced as a random factor to randomize the fixed CRPs with almost negligible extra cost. Because MFA are spatio-temporal uniqueness, it is difficult for attackers to predict and masquerade as legitimate devices. Based on the channel reciprocity of the legitimate transceiver, the randomized CRPs can be easily recovered to authenticate the legitimate devices. Simulations have confirmed that our proposed scheme outperforms other benchmark solutions to resist MAML with higher accuracy and lower computation time.