Scholarlink

The rapid evolution of wireless communication systems emphasizes the need for efficient data transfer mechanisms, particularly considering cellular communication. For efficient data transfer, this study considers clustering cellular users and showcases the performance gains across different performance matrices. We create a cellular communication environment on MATLAB and use K-Medoids, Hierarchical Clustering with Ward's Linkage, and DBSCAN algorithms to cluster the users. Subsequently, we evaluate the performance of the system by considering user data rate, throughput, and spectral efficiency. We also propose an algorithm, based on signal strength measurements and optimized using different existing techniques. In this study, we utilize the Calinski-Harabasz criterion, Gap statistic, and Silhouette score to determine the optimal number of clusters, refining network performance through successive adjustments. A comprehensive comparative study of clustering algorithms shows consistent improvements across all performance parameters, underscoring their robustness in ensuring reliable communication. Notably, the proposed algorithm boosted the signal strength by approximately 15% which resulted in improvement of other parameters such as user data rate and throughput.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Cellular User Clustering: Comparing Clustering Algorithms and Enhancing Network Performance

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Kong Zi Yan,
  • Max,
  • Saad Aslam,
  • Anwar P. P. Abdul Majeed,
  • Wei Chen

摘要

The rapid evolution of wireless communication systems emphasizes the need for efficient data transfer mechanisms, particularly considering cellular communication. For efficient data transfer, this study considers clustering cellular users and showcases the performance gains across different performance matrices. We create a cellular communication environment on MATLAB and use K-Medoids, Hierarchical Clustering with Ward's Linkage, and DBSCAN algorithms to cluster the users. Subsequently, we evaluate the performance of the system by considering user data rate, throughput, and spectral efficiency. We also propose an algorithm, based on signal strength measurements and optimized using different existing techniques. In this study, we utilize the Calinski-Harabasz criterion, Gap statistic, and Silhouette score to determine the optimal number of clusters, refining network performance through successive adjustments. A comprehensive comparative study of clustering algorithms shows consistent improvements across all performance parameters, underscoring their robustness in ensuring reliable communication. Notably, the proposed algorithm boosted the signal strength by approximately 15% which resulted in improvement of other parameters such as user data rate and throughput.