Variation Between DDC and SCAMSMA for Clustering of Wireless MultipathWaves in Indoor and Semi-Urban Channel Scenarios

Abstract

The performance of Simultaneous Clustering and Model Selection Matrix Affinity (SCAMSMA) and Deep Divergence-Based Clustering (DDC) in clustering wireless multipaths generated by COST 2100 channel model (C2CM) is compared. Enhancing the accuracy of clustering multipaths is an open area of research which the clustering approaches try to improve. Jaccard index is used as the clustering validity metric of the clustering approaches. The results of the clustering approaches are compared using the analysis of variance (ANOVA) toolbox of MATLAB and displayed using the box plots. Results show that the cluster-wise Jaccard index is different between SCAMSMA and DDC for indoor scenarios, while the membership-wise Jaccard index is not. On the other hand, the cluster-wise Jaccard index is not different between the clustering approaches for semi-urban scenarios, while the membership-wise Jaccard index is a little different. The clustering approaches can be used in indoor scenarios based on accuracy.