Hybrid Beamforming for Secure Transmission of Massive MIMO UAV Communication Networks

The powerful line-of-sight (LoS) characteristic between unmanned aerial vehicles (UAV) and terrestrial nodes could be leveraged by large-scale antenna arrays to significantly increase transmission rates, while the problem of secure transmission is emerging. Compared with fully digital beamforming, hybrid beamforming (HBF) could meet a compromise between cost and performance, which is precisely optimized in this article to maximize the secrecy rate of massive multiple-input–multiple-output (MIMO) UAV communication network. At first, we propose a secure HBF framework based on generalized eigenvalue decomposition and singular value decomposition. After that, considering the difficulty in directly solving the HBF optimization problem, an iterative algorithm based on successive convex approximation is developed for perfectly known channel state information (CSI). Furthermore, while CSI is imperfect, a neural network is constructed and trained to learn the statistical characteristics of wireless channels and complete the design of hybrid beamformer and combiner simultaneously. Performance validation and beampattern of the proposed HBF methods are given by extensive simulation results, which demonstrate the proposed secure HBF framework could considerably improve secrecy rate, and the deep learning-based HBF method possesses a positive average secrecy rate even with unknown eavesdropper's CSI.