Search Mechanism In Unstructured P2p Systems

At present, society is an information society in which information can be shared infinitely. And only sharing information can fully play its potential value. Development of network technology makes information sharing become particularly convenient. Especially emergence of P2P network in recent years greatly improves efficiency on information sharing. Information sharing is also called reuse of resource. The most important advantage of P2P network is the high resource utilization.To sufficiently use resources in P2P network, we must first find required resources. That is to say, search is executed in P2P network. Currently, search is the core problem of P2P research field. This paper mainly studies search in unstructured P2P network and its specific contents are as follows:LightFlooding was a kind of search query-based and content pulled in unstructured P2P network. Aiming at the fault of its redundant query messages and long search delay to rare resources, this paper proposed a kind of hybrid search scheme which combined unstructured P2P search with bidirectional Chord ring. Based on the k-walker random walk, that scheme utilized the node heterogeneity to modify the query request-forward manner by considering the ability of neighbor nodes. Moreover, resources that couldn't be found by unstructured P2P search were searched along bidirectional Chord ring. Simulation results show that hybrid search has better performances on redundant query messages and search delay to rare resources compared to LightFlooding.Among the search query-based in unstructured P2P networks, node passiveness and unstructured overlay led to large numbers of queries and brought long search delay and high system load. In order to solve this problem, ASAP was put forward. ASAP was a content push method in which node published advertisements in advance to propagate their common contents. Meanwhile, nodes optionally stored their interested advertisements from the other nodes. Although it could effectively locate contents, the efficiency of its advertisement delivery and search wasn't the optimal. Therefore, the algorithm called super-node constrained random was put forward to send advertisements, which adopted the idea of node heterogeneity and network layer. Moreover, the search algorithm was optimized with the strategy of advertisement replication. Simulation results show that modified ASAP called MASAP outperforms ASAP on success rate, response time and cost of search.