Multi-path clustering method based on bisecting K-means clustering in wireless channel modeling

In order to rationally allocate multi-path component (MPC) of wireless channels to different clusters, a multi-path clustering algorithm based on Bisecting K-means clustering was proposed for millimeter-wave channels to solve the problem of partial optimization of traditional K-means clustering methods. The MPC clustering was implemented based on the cluster split and iteration computation strategy, when the Mahalanobis distance (MD) was regarded as the multipath component distance (MCD). The experiment measurement in indoor millimeter-wave channel was conducted to verify the effectiveness and feasibility of the proposed method. The experimental results show that the clustering result of proposed method is more rational than those of traditional K-means clustering methods, and the highly correlated MPCs are reasonably and uniquely allocated to the same cluster.