Clustering of Wireless Sensors based on Ultra-Wideband Geo-Regioning

This paper introduces a novel approach for the clustering of Ultra-Wideband nodes based on their channel impulse responses (CIRs) to a central unit. We propose to model the CIRs as realizations of a mixture density, where the individual component probability density functions represent the clusters. It is shown that the clustering can be simply achieved by the application of the expectation maximization algorithm. Furthermore, this work studies the impact of a-priori knowledge and the number of available CIR observations on the clustering performance, which is evaluated based on measured CIRs.