Research And Implementation Of Resources Search Algorithm Based On Unstructured P2P Network

In recent years, P2P network becomes a hot topic in computer network technology, which is overlay network based on the network application layer. Compared with the traditional client/server(C/S) model, P2P network outperforms in fault tolerance, cost, scalability and resource sources. One of the important issues in P2P network is how to find and locate the information resources quickly and efficiently, namely the network resources searching problems. Compared with structured P2P network, unstructured one owns the features of no strict specified topology, network nodes are randomly connected, loosely organized. As unstructured P2P network suits better in practical P2P network, most of the current P2P applications are based on unstructured P2P network model.This thesis focuses on search algorithm in unstructured P2P network and proposes an improved algorithm(MABIT) which is based on the index table. In MABIT algorithm, network nodes need to maintain two resource index tables, which are the index table itself and neighbor node index table. A node is given a prior consideration for the neighbor of other nodes when the actual network distance among them is closer. During the communications in network, when a closer node is found or resources in neighbors is updated, nodes update their neighbors and the neighbor resources index information. In MABIT algorithm, the query of resources index table is mainly executed by the nodes which are at the index radius, the nodes which are not at the index radius only forward querying messages.In order to increase the search success rate, A message forwarding technique based on resource correlation among nodes is introduced in MABIT algorithm. Firstly, the resources in the node can be represented as a suffix tree structure, secondly, the similarity among nodes can be calculated using vector space model. The ranking in neighbor node index table is determined by the similarity between node and the source node, when the similarity is bigger, the corresponding node sorts frontal. Therefore, when receives a query request, the node forwards the request message to k-th neighbor node directly which are sorted frontally in neighbor resource index table.In this thesis, OPNET network simulation software is used to evaluate MABIT algorithm, which is compared with four existed algorithms of unstructured P2P network, experimental results show that MABIT algorithm can effectively reduce network load, search delay and improve search success rate.