Radar Enhanced Multi-Armed Bandit for Rapid Beam Selection in Millimeter Wave Communications

The complexity of learning the optimal beam for millimeter wave (mmW) communications using multi-arm bandit (MAB) algorithms linearly scales with the number of beams resulting in high latency. We propose integrating radar with communication to accelerate MAB learning by searching only those beams where the radar detects a mobile scatterer. Simulations show that our proposed radar-enhanced MAB reduces the beam exploration time resulting in improved throughput.