P2p Network Search Algorithm Research

With the advancement of information technology and the development of Internet, there is more and more information resource online, so how to get valuable information on the enormous internet has become an increasingly concerned problem. As a rising network computing mode, P2P break the traditional C/S mode, it is self-organized and has good fault-tolerance ability and expansibility. Fortune magazine call it as one of the four technologies which will affect the future of Internet. Compared with server-based search, the searching network in allusion to local data based on P2P has many advantages. For example, it has good search performances and exact search result; furthermore, it does not need costly server.Currently, the search mechanisms of existent structured or unstructured P2P systems both have merits and demerits. Unstructured P2P networks adopt blind search mechanisms(like flooding), although it supports flexible search, it will generate exponentially increasing redundant messages as the P2P network grows which lead to low search efficiency and bad extensibility. While DHTbased structured P2P networks can provide good search efficiency and scalability by deploying identifier-based search mechanism, it will pay very high price to maintain fixed topology, moreover, it cannot well accommodate to highly dynamic circumstances and fail to support flexible multi-keyword search like unstructured P2Pnetworks do.In the thesis, we analyzed the search mechanism of P2P networks. In order to solve such problem in unstructured P2P networks as blind search, low efficiency and redundant messages, we presents an algorithm that groups nodes based on their position and chooses to forward the searching demand to practically contiguous nodes dynamically according to their historical searching feedback results. Through calculating neighbor nodes' degree of interest correlation, the algorithm adjusts neighbor nodes regularly. The analysis of the algorithm and the results of experiments show that, compared to flooding algorithm, this algorithm improves performance in terms of searching time reduction, as well as controls the amount of messages forwarded and the proportion of repeating visit nodes, so it improves the searching efficiency.