DQN-based Joint Adaptive Beamwidth Control and Beam Tracking for mmWave Communications

Abstract

This paper presents a joint adaptive beamwidth control and beam tracking algorithm for mobile millimeter-wave (mmWave) communications based on deep Q-network (DQN). When the mobile station is highly dynamic, it may deviate from the beamwidth, thereby causing less robust beam tracking. Thus, the proposed algorithm aims to satisfy both robustness and high beam gain by adjusting the beamwidth proportional to the mobility. In the proposed algorithm, the possible actions that DQN can select encompass both beamwidth control and beam tracking. Among these actions, DQN selects the action that maximizes the received signal strength. Throughout simulations, we compare the proposed algorithm with the recent beam tracking algorithm without adaptive beamwidth control. The results confirm the effectiveness of the proposed algorithm for various mobile dynamics, especially in dynamic mobile environments.