Robust 3D Beamforming for Secure UAV Communications by DAE

Unmanned aerial vehicles (UAVs) are considered to play vital roles in the Sixth Generation (6G) networks and beyond. However, the confidentiality of UAV communication is sensitive to security threats owing to the broadcast nature and principal line-of-sight channel conditions. This paper proposes a three-dimensional (3D) robust beamforming method for UAV systems in terms of physical layer security in order to improve secrecy performance. In particular, attempting to maximize the average secrecy rate (ASR) of the considered system, a precisely designed unsupervised artificial neural network based on Denoising AutoEncoder (DAE) is used to optimize the beamformer for confidential signal with the simultaneous existence of artificial noise (AN). Simulation results show that the proposed approach can achieve higher average secrecy and more focused beams than existing previous methods in the same scenarios.