Deep Reinforcement Learning Coordinated Receiver Beamforming for Millimeter-Wave Train-Ground Communications

As more and people choose high-speed rail (HSR) as a means of transportation for short trips, there is ever growing demand high quality multimedia services. With its rich spectrum resources, millimeter wave (mm-wave) communications can satisfy the network capacity requirements HSR. Also, it possible receivers (RXs) to be equipped with antenna arrays in mm-wave communication systems due wavelength. However, HSRs run speed, received signal power (RSP) varies rapidly over cell lowest at edge compared other locations. Consequently, necessary conduct research on RX beamforming HSR band improve signal. In this paper, we focus train-ground system. To RSP, propose an effective scheme based deep reinforcement learning (DRL), develop Q-network (DQN) algorithm train determine optimal beam direction purpose maximizing average RSP. Through extensive simulations, demonstrate that proposed has better performance than four baseline schemes terms RSP most positions railway.