Fast mmWave Beam Tracking with Angular Velocity Estimation for Cellular-Connected UAVs

Beam tracking is a promising technology in mmWave-enabled cellular-connected unmanned aerial vehicle (UAV) communications. However, conventional beam tracking schemes always incur a large training overhead, and it is difficult to determine the time duration of a training cycle due to the high mobility of UAVs, which is essential for improving the effective achievable rate (EAR). To address this issue, we first adopt angular velocity estimation to obtain the beam coherence time, which serves as the time duration of the training cycles. To further reduce the training overhead in each training cycle, we then design an adaptive beam tracking algorithm based on bandit learning, where the actions are taken based on the accuracy of the angular velocity estimation. If the estimation is not accurate, more beams will be swept in the next cycle. In this way, the beam misalignment incurred by estimation inaccuracy will largely alleviate. Thus, the EAR can be effectively improved with smaller training overhead. The simulation results demonstrate the superior performance of the proposed algorithm in terms of the training overhead and the EAR.