Physical Layer Authentication for Mobile Devices in WLAN Systems: An Autoencoder-Based Approach

Physical layer authentication (PLA) enhances wireless security by using wireless channel features like channel state information (CSI) to authenticate transmitters and detect adversaries. While machine learning (ML) has been applied to improve PLA, most methods require adversary data or assume a stationary environment, limiting real-world practicality. This paper proposes an autoencoder-based PLA framework that relies solely on legitimate users’ CSI to distinguish them from adversaries in dynamic wireless environments. Using a wireless local area network (WLAN) testbed (e.g., Wi-Fi) with mobile and stationary devices in both line-of-sight (LoS) and non-line-of-sight (NLoS) scenarios, experimental results show that the proposed method outperforms existing schemes in authentication accuracy under mobility conditions.