Resource Allocation in Uplink NOMA-IoT Networks: A Reinforcement-Learning Approach

Non-orthogonal multiple access (NOMA) exploits the potential of power domain to enhance connectivity for Internet Things (IoT). Due time-varying communication channels, dynamic user clustering is a promising method increase throughput NOMA-IoT networks. This article develops an intelligent resource allocation scheme uplink communications. To maximise average performance sum rates, this work designs efficient optimization approach based on two reinforcement learning algorithms, namely deep (DRL) and SARSA-learning. For light traffic, SARSA-learning used explore safest policy with low cost. heavy DRL handle traffic-introduced huge variables. With aid considered approach, addresses main problems fair in NOMA techniques: 1) allocating users dynamically 2) balancing blocks network traffic. We analytically demonstrate that rate convergence inversely proportional sizes. Numerical results show that: Compared optimal benchmark scheme, proposed algorithms have lower complexity acceptable accuracy NOMA-enabled IoT networks outperform conventional orthogonal terms system throughput.