DQN-Based Beam Tracking Using Reward-Aware Exploration Without Additional Signaling

초록

Millimeter-wave communication offers high data rates but suffers from severe propagation loss, motivating repeater-assisted transmission and beamforming. This letter proposes a deep Q-network (DQN)-based beam tracking method without additional signaling for base station (BS)-repeater-user equipment (UE) systems. The proposed method uses normalized uplink received power (RP) and reward-aware exploration (RAE) to maintain beam alignment without additional pilot signaling. RAE adaptively adjusts the exploration range according to the observed reward, while a continuous reward improves sensitivity to beam alignment changes. Simulation results demonstrate higher RP and achievable rates than benchmark schemes, approaching ideal beamforming performance.

키워드