Estimation of the Number of Clusters in Multipath Radio Channel Data Sets

This communication deals with the grouping of the multipath radio channel components. The estimation of the number of clusters that a given data set exhibits is an important step in channel modeling and parameterization. We present a performance assessment of several cluster validity indices using Monte Carlo simulations. Synthetic channels were generated by means of the extended Saleh-Valenzuela model and used as input to a clustering algorithm which outputs several solutions with different number of clusters. The cluster validity indices allow the selection of the best solution. Finally, an example is presented where the best performing indices were applied to an experimental data set.