Intelligent Beam Training for Millimeter-Wave Communications via Deep Reinforcement Learning

Abstract

Millimeter wave (mmwave) communication has attracted increasing attention owing to its abundant spectrum resource. The short wave-length of mmwave signals facilitates exploiting large antenna arrays to achieve large array gains and combat large path-loss. However, the use of large antenna arrays and narrow beams leads to a large overhead in beam training for obtaining channel state information, especially in dynamic environments. To reduce the overhead of beam training, in this paper we propose an environment sensing based beam training algorithm via deep reinforcement learning. The proposed algorithm can sense the change of the environment and learn required latent probability information from the environment, and intelligently trains beams with a low overhead. In addition, the proposed algorithm does not require any priori knowledge of dynamic channel modeling, and thus is applicable to a variety of complicated scenarios. Simulation results demonstrate the effectiveness and superiority of the proposed intelligent beam training algorithm.