Energy-Efficient Multiple Access Scheme with Power Control for mMTC Networks.

The grant-free non-orthogonal multiple access (GF-NOMA) is especially conducive to support massive machine-type communication (mMTC) service. Owing to the limited energy resources and the reliability requirements for MTC devices, how to design an energy-efficient multiple access scheme is an urgent problem. In this paper, we investigate a comprehensive learning framework for GF-NOMA with power control scheme in mMTC networks, focusing on the power level design and the collision resolution. We formulate an optimization problem aiming to minimize the long-term system power consumption with satisfying the constraint of packet error rate. A scheme based on a deep reinforcement learning approach is further applied, comprising two interrelated subalgorithm, to optimize the power levels and the contention-transmission unit (CTU) allocation. The numerical results demonstrate the superiority of our proposed scheme over other resource allocation schemes, particularly in terms of enhancing the network energy efficiency.