Federated Learning Scheme for Physical Layer Authentication in Industrial IoT Application

With the recent advancements in communication technology and the advent of Internet of Things (IoT) devices, developing lightweight authentication methods for IoT systems is an essential requirement as these devices are resource-constrained. The new directions are towards using physical layer characteristics for authentication. With the advent of artificial intelligence, the task of authentication has become a significant burden on the authenticating device. In this paper, the use of federated learning to learn the physical layer attributes to provide lightweight collaborative authentication has been investigated. In this research, a measured channel state information dataset is used to evaluate the effectiveness of the proposed model based collaborative physical layer authentication approach on authentication accuracy. Although the CSI dataset represents a harsh industrial environment, the federated learning based physical layer authentication approach achieves more than 90% of authentication accuracy for various scenarios investigated in this research.