基于K均值聚类算法的NOMA用户分组

User Grouping for NOMA Based on K-Means Clustering Algorithm

摘要: 为解决非正交多址接入(Non-Orthogonal Multiple Access, NOMA)系统中的用户分组问题，提升系统吞吐量与用户公平性，本研究提出了一种基于K均值聚类算法(K-Means Algorithm, KMA)的用户分组方法。该方法以用户的信道增益及其对数信道增益作为聚类特征，旨在构建组内信道差异显著的用户分组，以优化NOMA的功率分配与干扰管理，并与分数功率分配(Fractional Transmit Power Allocation, FTPA)算法相结合。为评估所提算法性能，将其与基于信道差异的分组、基于距离的分组及随机分组等方法进行了对比分析，对比指标包括系统吞吐量、用户公平性、算法运行时间及用户速率分布。仿真结果表明，基于K-means的分组策略在系统吞吐量与用户公平性之间取得了更优的平衡，其性能显著优于随机分组和基于距离的分组方法，同时与基于信道差异的分组方法相比展现了可比或更优的综合性能。此外，可视化结果清晰展示了K-means算法依据信道特性所形成的用户聚类分布。该研究为NOMA系统中高效的用户分组提供了一种有效的解决方案。

Abstract: To address the user grouping problem in Non-Orthogonal Multiple Access (NOMA) systems and improve system throughput and user fairness, this study proposes a user grouping method based on the K-Means Algorithm (KMA). The method takes users’ channel gains and their logarithmic channel gains as clustering features, aiming to construct user groups with significant intra-group channel differences to optimize power allocation and interference management in NOMA, in combination with the Fractional Transmit Power Allocation (FTPA) algorithm. To evaluate the performance of the proposed algorithm, it is compared with grouping methods based on channel differences, distance-based grouping, and random grouping. Comparative metrics include system throughput, user fairness, algorithm runtime, and user rate distribution. Simulation results show that the K-means-based grouping strategy achieves a better balance between system throughput and user fairness, significantly outperforming random grouping and distance-based grouping methods, while demonstrating comparable or superior overall performance compared to channel-difference-based grouping. Furthermore, visualization results clearly illustrate the user clustering distribution formed by the K-means algorithm based on channel characteristics. This study provides an effective solution for efficient user grouping in NOMA systems.