Clustering Validation for mmWave Multipath Components in Outdoor Transmissions

Abstract

Radio channel propagation models for the mmWave spectrum are of paramount importance for the design and planning of future 5G wireless communications systems. Since transmitted radio signals are received as clusters of multipath rays, the problem arise about how to identify them, which is functional to extract better spatial and temporal characteristics of the mmWave channel. This paper deals with the validation of the results produced by the clustering process. Specifically, we estimate the effectiveness of the k-means clustering algorithm in predicting the number of clusters by using cluster validity indices (CVIs) and score fusion techniques. We consider directive transmissions in outdoor scenarios and we show the importance of the correct estimation of the number of clusters for the mmWave radio channel simulated with a software ray-tracer tool. Our investigation shows that clustering is no trivial task because the optimal number of clusters is not always given by one or by a combination of more CVIs. In fact, a few of the CVIs used in our study were not capable to determine correct partitioning. However, using score fusion methods and additional techniques we find two solutions for the number of clusters based on power and time of arrival of the multipath rays or based on their angle of arrival.