SMBSDD: An Improved Search Mechanism In Unstructured P2P Networks

Peer-to-Peer (P2P) networks are overlay networks on top of the Internet. Network nodes in P2P networks are equal to each other and can serve as both client and server. With a certain search mechanism, a network node locates and gets shared files through a direct connection from file-owning network nodes. Unstructured P2P networks are the most popular P2P networks and have been widely applied in areas of file-sharing, cooperative work, distributed computing and so on. With the problems of lacking of global topology, network nodes joining and leaving the network at will and shared files of network nodes could be anywhere regardless of the topology, the unstructured P2P network has a strong demand for search mechanism. How to locate files being queried faster and meanwhile reduce the communication costs during the Proceedingsess has been discussed a lot by researches of P2P network technologies.This paper introduced related background researches about P2P networks, Information Retrieval (IR) and Network Distance Prediction,researched the Flooding search mechanism in Gnutella 0.4 protocol as well as other search mechanisms in unstructured P2P networks. Considering the pure P2P network's characteristics of global-topology lacking and dynamic evolving as well as the Vector Space Model (VSM), we presented an improved search mechanism named SMBSDD (Search Mechanism Based on Similarity, Degree and Distance-Prediction) on the basis of the research. We grouped network nodes according to their interest vectors and introduced in the elements of node's historical successful query vector,node's degree and network distance-prediction to direct and forward the Query message to proper nodes.Then we discussed the feasibility of SMBSDD, illustrated, designed and realized the simulation system of SMBSDD. With the simulation experiments, we evaluated some technical indexes of SMBSDD, such as the number of Query message,average number of hops,success rate of Query and so on. The experiment results proved the effectiveness of SMBSDD.