With the evolution of 5G networks, optimizing resource allocation has become crucial to meeting the increasing demand for massive connectivity and high throughput. Combining Non-Orthogonal Multiple Access (NOMA) and massive Multi-Input Multi-Output (MIMO) enhances spectral efficiency, power efficiency, and device connectivity. However, deploying MIMO-NOMA in dense networks poses challenges in managing interference and optimizing power allocation while ensuring that the Signal-to-Interference-plus-Noise Ratio (SINR) meets required thresholds. Unlike previous studies that analyze user clustering and power allocation techniques under simplified assumptions, this work provides a comparative evaluation of multiple clustering and allocation strategies under identical spatially correlated network conditions. We focus on maximizing the number of served users under a given Quality of Service (QoS) constraint rather than the conventional sum-rate maximization approach. Additionally, we consider spatial correlation in user grouping, a factor often overlooked despite its importance in mitigating intra-cluster interference. We evaluate clustering algorithms, including user pairing, random clustering, Correlation Iterative Clustering Algorithm (CIA), K-means++-based User Clustering (KUC), and Grey Wolf Optimizer-based clustering (GWO), in a downlink spatially correlated MIMO-NOMA environment. Numerical results demonstrate that the GWO-based clustering algorithm achieves superior energy efficiency while maintaining scalability, whereas CIA effectively maximizes the number of served users. These findings provide valuable insights for designing MIMO-NOMA systems that optimize resource allocation in next-generation wireless networks.

翻译：随着5G网络的发展，优化资源分配已成为满足海量连接和高吞吐量需求的关键。非正交多址接入（NOMA）与大规模多输入多输出（MIMO）技术的结合提升了频谱效率、功率效率和设备连接能力。然而，在密集网络中部署MIMO-NOMA系统面临着管理干扰和优化功率分配的挑战，同时需要确保信号与干扰加噪声比（SINR）满足既定阈值。与以往在简化假设下分析用户聚类和功率分配技术的研究不同，本文在相同的空间相关网络条件下，对多种聚类和分配策略进行了对比评估。我们关注在给定服务质量（QoS）约束下最大化服务用户数量，而非传统的和速率最大化方法。此外，我们在用户分组中考虑了空间相关性——这一因素尽管对抑制簇内干扰至关重要，却常被忽视。我们在下行链路空间相关MIMO-NOMA环境中评估了多种聚类算法，包括用户配对、随机聚类、相关性迭代聚类算法（CIA）、基于K-means++的用户聚类（KUC）以及基于灰狼优化器的聚类（GWO）。数值结果表明，基于GWO的聚类算法在保持可扩展性的同时实现了更优的能量效率，而CIA则能有效最大化服务用户数量。这些发现为设计能优化下一代无线网络资源分配的MIMO-NOMA系统提供了重要参考。