Eliminating Clustering in the Propagation Tree of Semantic Peer-to-Peer Networks

B. Forstner, R. Kereskényi, and H. Charaf (Hungary)

Keywords

PeertoPeer Networks, Protocols, Topology, Modeling

Abstract

Some semantic approaches in the Peer-to-Peer information retrieval systems extends the connections of the nodes with promosing links. However, this extension causes clustering in the query propagation tree, which decreases the number of reached nodes and the efficiency of the retrieval. Moreover, the clustering causes unnecessary network traffic and redundant information processing at the nodes. In this paper we will propose a more optimal network topology for such semantic Peer-to-Peer systems that eliminiates clustering in the propagation tree. To be able to analytically describe the clustering effect, we will present a model that is aware of the clustering through an appropriate coefficient that shows the influence of the counterproductive links in the propagation tree. With the new topology, the number of messages required for a successful query can significantly be decreased, while the hit rate is increased by the semantic protocol. The low number of messages makes the protocols feasible in mobile Peer-to-Peer networks. We will validate our results with a series of simulation.

Important Links:



Go Back