The Research On An Improved Search Algorithm In Non-structured P2P Network

Peer-to-Peer network has received more and more attention by its good fault-tolerant, auto-organized and the resources sharing extendibility. The non-structured P2P network technique has already become one of the most popular and important applied techniques in the internet in recent years. Facing the complicated network environment, the existent non-structured P2P network search algorithms have limitation in searching for cost and performance. So they can not satisfy the request for application at present very well. According to the characteristic of the non-structured P2P network, how to establish an effective resource discovery mechanism, and seek one kind of high effective non-structured P2P network search algorithm to satisfy the application request of current complicated network environment has very good theory and practical significance.Based on the comprehensive research on the non-structured P2P network search technique, combined with the merit of learning-based P2P network search algorithm as well as the characteristic of the non-structured P2P network, an improved Learning-Based P2P network search algorithm is presented. The overleaped quantity of shared file denotes the similarity of user interest. The algorithm take some nodes that has the high hit probability within a period of time in the network as friend nodes, compose them to different groups using similarity rule and consider every code transmitting information as a router at the same time. By transmitting information in the interest groups established in every node, it makes independence nodes compartmentalized in logic, so it simplifies the resource searching.By simulating P2P network environment, the change of network cost and the rate of successful search when the algorithm has stabilized is tested and the comparison with the original algorithm is presented. The experimental result indicates better performance in the new study and classification mechanism, which can reduce the hunting zone effectively and reduce current capacity of network at the same time. The research has made beneficial attempt for achieving high effective and feasible P2P network search engine system.