6G Drones Predictive Channel Characterization Incorporating 3-D Aeronautic Random Trajectory

Unmanned aerial vehicle (UAV) communication channel research is an essential cornerstone for sixth-generation (6G) communication systems. Traditional UAV channel models lack accurate predictive capabilities and assume the flight trajectory to be pre-set. Motivated by this limitation, this letter firstly conducts 6G UAV predictive channel characterization incorporating three-dimensional (3-D) aeronautic random trajectory, which can capture relevant channel information in advance. Through the acquired ray tracing (RT) channel data validated by channel measurements, we derive spatial-temporal channel characterization for UAV air-to-ground multiple-input multiple-output (MIMO) communications. By leveraging the advantages of deep neural network (DNN), long short-term memory (LSTM) and self-attention mechanisms, this letter proposes and implements a predictive channel characterization model. The results show that the proposed model exhibits high accuracy and strong generalizability, and can be applied to scenarios where UAVs fly with 3D arbitrary trajectories.