A novel spatial beam training strategy for mmWave UAV communications

Abstract

Millimeter wave (mmWave) communications based on unmanned aerial vehicles (UAVs) are very promising in the fifth generation (5G) and future communication networks due to its ultra-high capacity. Considering the navigation of UAVs and narrow beam of mmWave, unstable beam pointing in the three-dimensional (3D) space becomes a key challenge for the mmWave UAV communications. In this paper, a novel spatial beam training strategy is proposed. The new method originates from the traditional two-dimensional (2D) hierarchical beam searching models, but takes the realistic 3D space of UAV flights into account. Firstly, a new 3D beam training model for UAV communication scenarios is developed. Meanwhile, the inverse discrete-space Fourier transform is used to generate the hierarchical codebook to construct the training beam with flat-topped characteristic and different beam width. In addition, a greedy geometric (GG) algorithm is proposed to generate the optimal beam to deal with the practical constraints of mmWave transceivers. The simulation results demonstrate that our proposed spatial beam training method can effectively improve the accuracy and efficiency of beam selection in 3D space environment.