Flexible Density-based Multipath Component Clustering Utilizing Ground Truth Pose

Accurate statistical characterization of electromagnetic propagation is necessary for the design and deployment of radio systems. State-of-the-art channel models such as the Enhanced COST 2100 Channel Model utilize the concept of clusters of multipath components, and characterize channels by their inter- and intra-cluster statistics. Automatic clustering algorithms have been proposed in literature, but the subjective nature of the problem precludes any from being deemed objectively correct. In this paper, a new algorithm is proposed, based on density-reachability and ground truth receiver pose, with the explicit focus of extracting clusters for the purpose of channel characterization. Measurements of downlink signals from a commercial LTE base station by a passenger vehicle driving in an urban environment with a massive antenna array on the roof are used to evaluate the repeatability and intuitiveness of the proposed clustering algorithm.