Research Of P2P Search Based On Structured Overlay Network

There are huge quantity information resources in the internet, but the traditional search engines based on network crawler can only retrieve a tiny fraction of them, moreover, the traditional search engines are not good at searching the information that existed in the form of dynamic web pages or stored in the network edge, but the new generation of search technology that based on Peer-to-Peer (P2P) thought may extend the information retrieval service to these places. P2P search technology has good real-time and validity, and its cost is low, its search depth and breadth are hard to be reach by traditional search technology.In this paper, we firstly introduce some knowledge and theory that related to P2P search and then discuss the principles and characteristics of present P2P search technologies. After that, we analyze the challenges that partition-by-keyword (PBK) faces in large-scale P2P network, and then present a new P2P search technology called partition-by-zone (PBZ) which based on structured network and using special index partitioning strategy that combine term partitioning and document partitioning. We describe PBZ's zoning rules, data locating algorithm, index building and maintaining algorithm, search algorithm and adaptive partition algorithm, and then make a theoretical analysis on its load balancing, network bandwidth cost, search delay and robustness. And then we introduce the architecture and implementation of a search engine using PBZ. Finally we verify PBZ's feasibility and excellent performance through several simulations.