Beam Domain Channel Modeling and Prediction for UAV Communications

Due to the agile three-dimensional (3D) mobility and flexibility, unmanned aerial vehicles (UAVs) have shown great promise for on-demand communications in sixth-generation (6G) wireless networks. In UAV communication systems, 3D beamforming is an effective technique for performance enhancement. To take full advantage of 3D beamforming in UAV communication systems, we need to accurately characterize and efficiently predict highly dynamic UAV channels in the beam domain. This paper first proposes a novel beam domain channel model (BDCM) considering random UAV trajectories and fuselage vibrations. Then, we propose a beam domain channel tracking algorithm to capture variations of multipath component (MPC) parameters in UAV channels. Finally, we put forward a novel beam domain channel prediction scheme for UAV communication systems utilizing channel sparsity and high temporal correlation in the beam domain. The proposed channel prediction scheme can extract channel variation trends utilizing the echo state network (ESN) and predict channel parameters in subsequent time blocks based on the history channel information. Simulation results show that the proposed channel prediction scheme outperforms the conventional prediction scheme based on angular speed estimation in terms of both prediction accuracy and communication system performance.