A density-based approach to node clustering in decentralized peer-to- peer networks
Qingwei Shi, Zheng Zhao, Hu Bao · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2007
Efficient organization of the nodes in decentralized peer-to-peer (P2P) networks is a challenging problem, especially in the absence of a global schema. Node clustering is an available way to optimize infrastructure and decrease traffic cost in P2P networks. This paper proposes a Density-based Distributed Node Clustering (DDNC) approach to discovering clusters in P2P networks. This approach is completely distributed, in which each node only depends on the knowledge of its neighbors for node clustering. Unlike other graph based algorithms, the DDNC approach utilizes density of node's neighbor for discovering clusters. For a given node, the DDNC determines its neighbor density by computing the link time with its neighbors, which not only considers the node connectivity but also connection quality. The DDNC scheme can also dynamically adapt its clusters according to the participation and departure of nodes. Experimental results have shown ours scheme's feasibility and efficiency.