AN EFFICIENT ALGORITHM FOR CLUSTERING NODES, CLASSIFYING AND REPLICATION OF CONTENT ON DEMAND BASIS FOR CONTENT DISTRIBUTION IN P2P OVERLAY NETWORKS

Abstract

This paper proposes an efficient algorithm for clustering nodes, classifying and replication of content on demand basis for content distribution in p2p overlay networks. Peer-to-peer overlay networks are a good solution for distributed computing than client-server model since nodes in P2P networks act both as client and server. With such functionality of managing replicas is an astonishing task and we designed an architecture for content distribution. In the proposed architecture, nodes are grouped into strong, medium and weak clusters based on their weight vector. More copies of Class I content are replicated in strong clusters that are having high weight values. Class II and Class III clusters consumes medium and low queries. Routing is performed hierarchically by broadcasting the query to the strong clusters first, then to medium clusters and finally to weak clusters. From extensive simulations using NS2 simulator, the proposed architecture achieves less bandwidth consumption, reduced latency, reduced maintenance cost, strong connectivity and query coverage.