D2D Assisted Q-Learning Random Access for NOMA-Based MTC Networks

Machine-type communications (MTC) should account for half the connections to internet by 2030. The use case massive MTC (mMTC) allows applications connect a number of low-power and low-complexity devices, leading challenges in resource allocation. Not only that, mMTC networks suffer under rigid random access schemes due ultra-dense nature resulting poor performance. In this sense, paper proposes $Q$ -Learning-based method machine-type communications, with device clustering non-orthogonal multiple (NOMA). traditional NOMA implementation increases spectral efficiency, but at same time, demands larger -Table, thus slowing down convergence, which is known be highly detrimental effect on networks. We pre-clustering through short-range device-to-device technology mitigate drawback, allowing devices operate smaller -Table. Furthermore, previous selection partner us implement full-feedback-based reward mechanism so that clusters avoid time slots already successfully allocated. Additionally, cope negative impact system overload, we propose an adaptive frame size algorithm run base station (BS). It adjusting network load, preventing idle underloaded scenario, providing extra when overloaded. results show great benefits terms throughput proposed method. addition, clusters, as well adaptation, are analyzed.