Online Learning-Based Beamwidth Optimization for Initial Access in Millimeter Wave Cellular Networks

The use of highly directional antennas in millimeter wave (mmWave) cellular networks necessitates precise beam alignment between a base station (BS) and a user equipment (UE), which requires beam sweeping over a large number of directions and causes high initial access (IA) delay. Intuitively, wider beams could lower this delay by requiring fewer sweeping directions. However, this results in a weaker received signal and a higher risk of misdetection, which potentially increases the expected IA delay by requiring more rounds of sweeping to discover a UE. In this paper, we propose a beamwidth optimization framework for both single-link and dual-link mmWave cellular networks, aiming to minimize the beam sweeping delay for a successful IA. We first analyze the impact of beamwidth on misdetection probability and formulate the beamwidth optimization problem accordingly. Then, we present the beam sweeping protocols that support beamwidth optimization. After that, we formulate the beamwidth optimization problem based on the multi-armed bandit framework and propose an online learning-based solution. Simulation results show that the proposed solutions can decrease the beam sweeping delay by more than 50% compared to the benchmark schemes.