The Study Of Resource Location Key Techniques In P2P

Since the P2P computing emerged, P2P technology has been developed for14years, and become the most important application on the Internet. With the popularity of the P2P application and quick growth of P2P users, P2P network has been presented huge amount, decentralized, dynamic characteristSAC. Hence, how to effectively locate the resource in huge P2P network is one of the key challenges needs to solve in the field of P2P. The researches found that optimizing the topology, improving search algorithms and changing data organization are the three effective way to improve search, Therefore, in order to improve the search efficiency of the system, the topology building, message routing and resource duplicating are three aspects to improve. The improvement of resource location efficiency still faces three major problems:1) uneven connections of super nodes in the topology optimization and super nodes' service capabilities;2) assessment and search of interest distance for nodes;3) the copy selection in nodes set, setting duplication on the nodes less visited leads to the utilization of a copy.This paper, under the support of the scientific and technological projects in Wuhan namely'Wuhan City public service platform for the talents', studies the optimizing techniques for the resource location in the P2P network from three aspects, where low service capabilities caused by Lower super-node connections, interest degree evaluation and search efficiency and the utilization rate of a resource copy. First this paper introduces the P2P network resource location research background, significance and the current study status, and then proposed hierarchical network topology management technology, interest-based search technology and the file access rate based copy replication protocol to solve the aforementioned questions. In this paper, the contributions and innovations in theoretical model algorithms and practice are as followed:l)Hierarchical P2P network topology building protocol STMSHierarchical P2P networks are proposed to improve the search efficiency, load balance and robustness of the network. The existing topology structure building protocols thought that as long as the node capacity is large enough and then it could be used as a super node. This ignores the heterogeneity of the nodes on the number of connections, resulting in the small number of leaf nodes connected to the selected super node and the low clustering coefficient. For the super node construction strategies represented by SG-2, the reduction of the super nodes'clustering coefficient is up to30%, leading to the excessive super nodes and low service rate. Therefore, this paper studies the super node selection with delay constraint. Preferentially select nodes with large enough capacity and the largest number of connections meeting the delay threshold as super nodes, to remove large capacity nodes with a small number of connections, and proposed topology management protocol STMS. Experiments show that under the same network environment, STMS compared with the optimal topology configuration protocol STMS can improve the clustering coefficient by15%-20%and decrease super nodes and leaf nodes proportion by8%-10%.2)Search protocol based on the interest on the P2P, IBSOn P2P network, when selecting the nodes with closest interest distance to forward, the selected nodes may have different interest. It leads to the inefficient search. Measurement of interest of existing methods entirely depends on the hit rate of the node, which usually is not true. To this end, this paper proposes a node interest model DUI, which could estimate the possibility of sharing fi after sharing fi in one node using the past clues, so it can reflect the real network environment. On this basis, an interest based search algorithm IBS is proposed. Experiments show that under the same network environment, IBS improves search success rate higher by10%than SOSPNET.(3) Self-adapted copy replication strategy based on visiting amount FTRCopy replication technique is to improve the reliability and availability of the data by increasing the redundancy in the system. Existing algorithms only uses the visiting amount to select the replication destination nodes, which could appear the 'bumps'phenomenon and cause lower replication utilization. This paper presents an adaptive replication management strategy FTR. FTR has two advantages:the FTR selects destination nodes by their both bandwidth and capacity, further taking into account the influence on other nodes caused by the destination node. Because the query encounters these copies and hits, query will not be forwarded to the downstream nodes, which will lead to the query reduction in the flow. It's necessary to reassess flow of the unselected node to ensure high copy hit rate. Compared to EAD algorithm, our method acquires high hit rate and the better efficiency, meanwhile greatly reduces the'bumps', and takes full advantage of a copy of a document. Experiments show that in the same network conditions, compared with the EAD, FTR copy utilization is increased by10%and the success rate is increased by15%.In summary, our works fully study the characteristSAC of P2P network and resource location technology, and explore P2P network topology structure building, message routing and copy replication, which is greatly meaningful for the large scale P2P applications. Finally, summarize the results and make prospects.