Selective Flooding Based on Relevant Nearest-Neighbor using Query Feedback and Similarity across Unstructured Peer-to-Peer Networks

Abstract

Problem statement: Efficient searching is a fundamental problem for unstructured peer to peer networks. Flooding requires a lot of resources in the network and thus will increase the search cost. Searching approach that utilizes minimum network resources is required to produce efficient searching in the robust and dynamic peer-to-peer network. Approach: This study addressed the need for efficient flood-based searching in unstructured peer-to-peer network by considering the content of query and only selecting peers that were most related to the query given. We used minimum information to perform efficient peer selection by utilizing the past queries data and the query message. We exploited the nearest-neighbor concept on our query similarity and query hits space metrics for selecting the most relevant peers for efficient searching. Results: As demonstrated by extensive simulations, our searching scheme achieved better retrieval and low messages consumption. Conclusion: This study suggested that, in an unstructured peer-to-peer network, flooding that was based on the selection of relevant peers, can improve searching efficiency.