Joint Channel Selection and Power Control for NOMA: A Multi-Armed Bandit Approach

The scarcity of the resources and the steady increasing number of connected devices have spotlighted Non-Orthogonal Multiple Access (NOMA) techniques in order to enable a massive connectivity, where the receiver use Successive Interference Cancellation (SIC) in order to separate multiple users’ signal. In this paper, we focus on the uplink transmission in power domain NOMA. We investigate the joint resource allocation and power control problem, and we propose a solution based on Multi-Armed Bandit (MAB). Furthermore, we investigate the exploration-exploitation tradeoff in the proposed reinforcement learning to allow users to choose the appropriate resource block and power level in a distributed manner. The resulting scheme yields better performance compared to the NM-ALOHA and shows an improvement in terms of throughput efficiency.