Proportional Fair Decentralized Scheduling for mmWave D2D Communications

The symbiosis between millimeter-wave (mmWave) and device-to-device (D2D) communications has attracted considerable attention in the last years. This traction is due to the potential of mmWave to circumvent the spectrum shortage in the sub-6GHz frequencies. Further, directional communications used in mmWave significantly reduce interference and enhance the spatial sharing gain. As a result, mmWave can assist high-density mmWave D2D networks. Despite the common belief that the operation regime of mmWave transmissions is noise-limited due to the use of directional antennas, it has been recently shown that mmWave's operation transitions between a noise-limited and an interference-limited regime depending on the network density. The different regimes call for a distinct channel access mechanism. While some solutions have been proposed, a one-fits-all MAC solution is still missing. To this end, we propose the first decentralized MAC scheduling mechanism that adapts to the interference conditions and copes with network dynamics. We show that it almost surely converges in finite time, and it provides proportional fairness upon convergence. Further, it can achieve up to 100% channel efficiency in both regimes within a short convergence time.