Automatic Clustering of MIMO Channel Parameters using the Multi-Path Component Distance Measure

This paper addresses the problem of identifying clusters from MIMO measurement data. Conventionally, visual inspection has been used for the cluster identification, however this approach is impractical for a large amount of measurement data. Moreover, visual methods lack an accu- rate definition of a "cluster" itself. We propose to use a previously introduced metric, the multi- path component distance (MCD), to calculate the distance between single multi-path components (MPCs) estimated by a channel parameter estimator, such as SAGE. The metric scales the different dimensions of the data to be in the in- terval of (0...1) and also solves the problem of the angular periodicity. We implemented this metric in the well-known hierarchical tree clustering algorithm. To assess the performance improvement of the new met- ric, the clustering algorithm is subsequently applied on syn- thetic data generated by the 3GPP spatial channel model (SCM) using the MCD, the well-known euclidean distance metric, and the joint squared euclidean distance as distance functions. Finally we verify the applicability of the met- ric by results from clustering real-world measurement data from an indoor big hall environment.