FCM-based clustering performance analysis in network topology

This paper compared the clustering performance of the proposed FCM-based clustering scheme with HCM to the hierarchical topology optimization in overlay multicast. The paper firstly depicts the proposed scheme including 1) defining the clustering feature distance based on overlay path latency, 2) calculating the Euclidean distance within each pair of overlay nodes to construct the fuzzy membership matrix, and 3) dynamically adapting the fuzzy clustering threshold. Then it designed the experiments to analyze the clustering performance with HCM. The experiments illustrated that the FCM-based clustering scheme achieves better clustering performance for the topology optimization.