Adaptive learning based directional MAC protocol for millimeter wave (mmWave) wireless networks

Directional antennas are used to counter the increased path loss at millimeter wave (mmWave) frequencies (beyond 30 GHz). Usage of directional antennas imply lack of channel sensing on the part of various network nodes. Consequently coordination among various network nodes becomes critical for efficient network operation. In this paper we propose the adaptive learning directional medium access control (AL-DMAC) protocol based on reinforcement learning, to facilitate implicit coordination among various nodes. Our simulation results demonstrate that the AL-DMAC protocol along with a suitably chosen neighbor discovery mechanism yields higher network throughput than both the naive directional slotted ALOHA (DSA) protocol and the memory guided directional medium access control (MD-MAC) protocol. Further the AL-DMAC protocol and the MD-MAC protocol have comparable performance in terms of fairness, measured in terms of the Jain's fairness index.