Research on the Search Scheme for Rare Items in Unstructured P2P Networks

Search method is the key technique in unstructured P2P systems. The existing search methods are efficient for locating popular items, but not so for locating rare items. Therefore increasing the search efficiency for rare items would improve the network availability and user experience. In this paper, we propose a novel search scheme for rare items. First, we present a data structure of TDBF for counting the files' local popularities by using Bloom Filter data compressing technique. Second, a replication strategy and a dynamic search strategy for rare items are proposed, which could make fully use of the heterogeneity of the P2P networks and also take the load balance into account. We conduct comprehensive simulations to evaluate the algorithm. The simulation results show that our approach could significantly improve the search quality of the rare items as well as reduce system traffic overhead. Index Terms - P2P, rare item, replica, Bloom Filter