Reinforcement Learning-Based Joint Beamwidth and Beam Alignment Interval Optimization in V2I Communications

The directional antenna combined with beamforming is one of the attractive solutions to accommodate high data rate applications in 5G vehicle communications. However, the directional nature of beamforming requires beam alignment between the transmitter and the receiver, which incurs significant signaling overhead. Hence, we need to find the optimal parameters for directional beamforming, i.e., the antenna beamwidth and beam alignment interval, that maximize the throughput, taking the beam alignment overhead into consideration. In this paper, we propose a reinforcement learning (RL)-based beamforming scheme in a vehicle-to-infrastructure system, where we jointly determine the antenna beamwidth and the beam alignment interval, taking into account the past and future rewards. The simulation results show that the proposed RL-based joint beamforming scheme outperforms conventional beamforming schemes in terms of the average throughput and the average link stability ratio.