A Dynamic Fuzzy Clustering Scheme in Overlay Multicast

This paper proposed a dynamic fuzzy clustering scheme with adaptive fuzzy threshold to the hierarchical topology optimization in overlay multicast. The scheme firstly normalized the two routing metrics - overlay path delay and overlay node access bandwidth-- to comprise the clustering feature vector, and then calculated the Euclidean distance within each pair of overlay nodes to construct the fuzzy membership matrix. Furthermore, inspired by the statistic histogram principle, the scheme dynamically adapted the fuzzy clustering threshold to the statistic results from the fuzzy membership matrix for the better clustering results. Compared with non-clustering method and HCM, the experiments illustrated that the proposed scheme performs better in both the clustering performance and the routing optimization.