Power-Profile in Q-Learning NOMA Random Access Protocols for Throughput Maximization

Power-Profile in Q-Learning NOMA Random Access Protocols for Throughput Maximization

연구 분야: Networking

학회: Journal of Network and Systems Management

With the introduction of 5 G technology, various communication services are becoming integrated into more sophisticated systems, such as Ultra-Reliable Low-Latency Communications (URLLC) and Massive Machine-Type Communications (mMTC). The goal is to facilitate seamless connectivity for many devices, ensuring minimal transmission delays. This advancement finds application in diverse fields like remote surgery and intelligent transportation systems. To achieve this, one approach involves the utilization of random access protocols. Implementing multi-power-level systems under Non-Orthogonal Multiple Access (NOMA) allows devices to efficiently share the same resource block (time and frequency) for transmission while varying their transmission power. This work delves into determining the optimal power distribution profiles in power domain NOMA-based random access networks. Specifically, it examines linear, quadratic, square-root, and exponential multi-power level profiles to identify the most efficient multi-power Q-learning random access strategy. Additionally, we investigate the optimal number of power levels for these profiles, considering factors such as latency and throughput. We analyze the impact of minimum SINR requirements on the overall system throughput.