The Study Of Searching Algorithm Based On P2P Network

Along with the development of the Internet and its increasing user number, Peer-to-Peer (P2P) network is attracting more and more attention in the world as a new network application mode. An increasing number of specialists have emerged into the study of it and have got many contributions. However, the research on unstructured P2P network is still few while most of the attention was focused on structured P2P network. According to the definition of Peer-to-Peer Working Group Committee, P2P can be used in the file sharing, distributed computing and so on. Gnutella networks model is the representative of pure P2P systems, but the peers and its shared files are searched and found by flooding in Gnutella. When new peers joining the P2P net, the number of searching messages as well as the latency flux generated by every message is increasing.To avoid a large volume of unnecessary network traffic incurred by the flooding-based search, many improved statistics-based search mechanisms have been proposed. Instead of flooding to all immediate overlay neighbors, a peer selects only a subset of its neighbors to query based on some statistics information and heuristic algorithms. Nevertheless, these search mechanisms only find some special peers and search files in those peers so that many available peers are ignored. The thesis researches those points and imports the dynamic route idea, and uses methods to initiatively get the network changes and store those in the route table. By using route table, the index size is proportional to the number of neighbors, rather than to the number of documents.