Deep Learning Driven 3D Robust Beamforming for Secure Communication of UAV Systems
Runze Dong, Buhong Wang, Kunrui Cao
Abstract
Beamforming is a promising technique to enhance the security of wireless transmission, while the optimal beamforming design with partial channel state informing (CSI) is challenging. This letter develops a three-dimensional (3D) robust beamforming method for unmanned aerial vehicle (UAV) communication systems in the physical layer security perspective. Specifically, aiming at maximizing the average secrecy rate of the considered system, a precisely designed neural network is trained to optimize the beamformer for confidential signal and artificial noise (AN), with partial CSIs of legitimate UAV and eavesdropping UAV. Simulation experiments show that the proposed deep learning (DL) based method could achieve better secrecy rate and flexible beam steering than benchmarks.