Studying On P2P Super Node Selecting Mechanism Based On Trust Model

In recent years, P2P technology is becoming a widely used new networkapplication models, and it use a variety of terminal equipment in the Internetenvironment for collaborative computing and resource sharing and otherapplications. With the rapid development of network technology and continuallyenhance of personal computers’ performance, the Internet computing model is agradual shift from the traditional C/S mode to the P2P mode.In actual P2P networks, however, the network dynamics and nodes’heterogeneity in the capacity of storage, computing, bandwidth and otherservices make management difficult. Besides, due to the anonymous,self-organizing and other features of P2P network, any node is free to join thenetwork, or access to resources and services. Also, because of dynamicallyjoining or exiting, nodes in P2P network can’t guarantee that all thecorresponding nodes are able to provide honest and reliable service and goodresources, even worse, some nodes provide malicious service.In order to improve the quality of P2P network management services, in thispaper, we introduce the concept of super-node by selecting one or two supernodes in a certain area, by which the network managing and providing servicesfor the common nodes of the region.Based on the P2P super nodes based on trust mode selection mechanism, Thispaper puts up comparative study on existing P2P super nodes selection methods.Our research follows:(1) Researching and analyzing four kinds of common P2P network structure:centralized directory networks, fully distributed unstructured network, fullydistributed structure network, the hybrid structure.(2) Studying and analyzing several common trust evaluation models: centralizedtrust evaluation model, recommended distributed trust evaluation model based onthe global and distributed trust recommended based on local assessment model.(3) Introducing the super nodes of the P2P network selection mechanism intotrust model. Using the total trust of nodes (direct trust and recommendation trustin the process of weighted average) as an indicator of the evaluation of the supernode; and in the step of calculating the trust, the award/penalty factor is introduced to punish the malicious service node, while providing bona fideservices to reward these nodes which services positively. This strategy canimprove nodes’ positivity of services.(4) Introducing of threshold filtering methods in the super-node selection process.If too much nodes participate in the selection of nodes in the network, comparingevery ordinary node in the region may significantly reduce the performance ofP2P networks. In order to avoid this kind of situation, in this article, before thesuper-node of the P2P network selection, we firstly filter the common node fromthe candidate super-node with threshold. In this step, ordinary nodes will beexcised out, which does not meet the basic conditions of the super node.(5)Taking the analytic hierarchy process in the super-node selection process toensure the super-node of the last selected is optimal.(6)Developing candidate super-node strategy to avoid the sudden departure ofthe super-node, which results the P2P network service interruptions.