Eavesdropper Avoidance through Adaptive Beam Management in SDR-Based MmWave Communications

High-frequency systems use beamforming to mitigate the increased path loss. As the resulting beams become highly directional, Millimeter Wave (mmWave) radios conduct a beam sweep to probe all possible angular directions to locate each other and establish communication. In this paper, we propose an adaptive beam management strategy that leverages beam sweeping to avoid eavesdroppers and other potential attackers. Our solution employs Deep Reinforcement Learning (DRL) to dynamically select a subset of beams in the transmitter codebook. We evaluate this solution through a proof-of-concept implementation using a combination of Software-Defined Radios (SDRs) and commercial mmWave equipment, and show the improvements in the secrecy capacity.