Deep Reinforcement Learning Framework for Efficient Resource Allocation in NOMA Systems

Abstract

The rapid growth of next-generation wireless networks has led to a substantial proliferation in the use of smart devices and emerging applications, which demand high computational power and low latency. Managing these devices within limited resources presents a substantial challenge. Non-Orthogonal Multiple Access (NOMA) has gained huge attention as a favourable technique to address these challenges by allowing multiple signals to be transmitted and received concurrently on the same frequency band, thereby improving spectral efficiency, network capacity, and energy efficiency. In this work, we focus on optimizing resource allocation in NOMA systems, particularly in the context of 5G networks. Despite the advantages of NOMA over traditional Orthogonal Multiple Access (OMA) methods, efficient dynamic resource allocation remains a critical challenge. To address this, we propose a novel model that leverages deep reinforcement learning for optimized resource allocation in NOMA systems. The proposed model demonstrates significant improvements in energy efficiency and spectral efficiency compared to existing methods, while also addressing challenges related to power allocation, user clustering, and interference management. Our approach provides a scalable and adaptive solution for enhancing the performance of NOMA-enabled 5G networks.