IJCTE 2024 Vol.16(1): 12-20
DOI: 10.7763/IJCTE.2024.V16.1349
Concurrent and Spectral Clustering of Wireless Waves
Jojo Blanza 1,* and Lawrence Materum 2,3
1. Electronics Engineering Department, University of Santo Tomas, Philippines
2. Department Electronics and Computer Engineering, De La Salle University, Philippines
3. International Centre, Tokyo City University, Tokyo, Japan
Email: jfblanza@ust.edu.ph (J.B.); materuml@dlsu.edu.ph (L.M.)
*Corresponding author
Manuscript received September 4, 2023; revised October 26, 2023; accepted December 6, 2023; published January 30, 2024
Abstract—Unsupervised clustering is generally used to identify groupings of wireless waves from estimated multipath parameters in order to optimize the cluster count and membership. The estimated parameters exclude cluster labels. For a better comparison of clustering approaches, cluster-based wireless channel models provide the labels. This work proposes that using the Three-Constraint Affinity Matrix (3CAM) in formulating the affinity or similarity matrix improves the clustering accuracy. Datasets generated from the European Cooperation in Science and Technology (COST) 2100 Channel Model (C2CM) were used and subjected to directional cosine and whitening transforms. Simultaneous Clustering and Model Selection Matrix Affinity (SCAMSMA), 3CAM-SCAMSMA, Spectral Clustering (SC), and 3CAM-SC were used to concurrently determine cluster count and membership. Various studies on multipath clustering give only the number of clusters. Others would state only the validity index of the membership of clusters. The problem with such an approach is that the correctness of the number of clusters is not an assurance that the membership of the clusters is accurate. The four clustering approaches solve this problem by determining the number of clusters and their membership. Thus, knowing each technique’s performance is essential. In the algorithms of all the clustering approaches, cluster count aims to ensure that the target cluster count is within the vicinity of the reference clusters. The cluster count and membership accuracy are computed through the cluster-wise Jaccard index of the multipath membership to their clusters. The performance of the clustering approaches was validated using the Jaccard index by comparing the calculated data with the reference data. The results show that 3CAM-SCAMSMA improved the clustering accuracy of SCAMSMA by an average of 5.218% in semi-urban scenarios. At the same time, 3CAM-SC increased the performance of SC in indoor scenarios by an average of 44%. 3CAM-SC is the most robust clustering approach, registering the highest accuracy and slightest variation.
Keywords—5G, channel model, clustering algorithms, data handling, Multiple-Input Multiple-Output (MIMO)
[PDF]
Cite: Jojo Blanza and Lawrence Materum, "Concurrent and Spectral Clustering of Wireless Waves," International Journal of Computer Theory and Engineering vol. 16, no. 1, pp. 12-20, 2024.
Copyright © 2024 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).
