A Novel Physical Layer Authentication Scheme of Internet of Things Devices Based on Center-Enhanced Open-Set Classifier

The development of the Internet of Things has provided a wide range of applications for consumer electronic devices. Due to impairments in devices hardware, radio frequency fingerprint (RFF) extracted from wireless signal can be used for physical layer authentication towards these devices, which has recently emerged as a promising approach for physical layer security. In response to the problem of complex channel environments and the inability to obtain illegitimate devices RF signal samples in advance, a novel physical layer authentication scheme is proposed for closed-set recognition and open-set recognition in this paper. Firstly, we extract RF robust feature combined amplitude and phase information from the received RF signal samples to mitigate wireless channel effects. Data augmentation strategy is presented to improve robustness. Then center-enhanced open-set classifier (CE-OSC) is introduced in the proposed scheme. Specifically, the loss function is adapted by using center loss function to facilitate Weibull model fitting in open-set classifier for the improvement of recognition accuracy. Finally, we carried out extensive experiments based on commercial long range radio(LoRa) devices under varying channel and openness conditions. The results have successfully demonstrated the outstanding performance of the proposed scheme compared with other schemes.