Research On The Trust Aware Resource Transaction Model In Peer-to-Peer Networks

With the development of computer and communication technologies, users have been storing more and more valuable contents in their own computers, and this has led to the change of contents storage center from the core of Internet to the end users. Thus, the exchange model of Internet contents has also changed from traditional Client/Server model to peer-to-peer model, that is, P2P network. Generally speaking, P2P network means the overlay network above the current Internet. Each peer in the network is both the service provider and the service demander, and also responsible for the routing of messages from other peers. P2P technologies are the research focuses of network academic community and the hot topics in industry.While P2P network has been widely deployed and successfully applied, there are still a lot of issues to study. Focusing on the issues caused by the autonomous and selfish features of peers, such as network dynamicity, peers' heterogeneity and unreliable resources, this dissertation discusses the solutions for P2P resource management based on the market mechanism and trust mechanism.Firstly, the dissertation introduces the concept, classification and applications of P2P network and deeply analyses the main reasons that cause network dynamicity, peers' heterogeneity and unreliable resources. By analyzing the merits and problems in P2P resource management with the market mechanism, the necessity and feasibility of introducing the trust mechanism are proofed. On this basis, this dissertation proposes the research idea on P2P resource management combined with the market mechanism and the trust mechanism.Secondly, this dissertation proposes the trust aware resource transaction model, abbreviated to Tart Model, and describes the design ideas, the market transaction model and the peer model. In Tart Model, trust is an important metric measuring the reputation of peers in the market transactions, which can not only stimulate peers to obey the transaction contract, but also provide an indirect penalty method for dishonest service providers.In chapter 3, based on summarizing the classification and problems of the current trust evaluation system, the dissertation brings forward a vectorial-similarity based service trust evaluation model recurring to the trust model of the society network. This model uses 'time-sensitive factor' to reflect the recent behavior features of peers in the evaluating results. Meanwhile, based on the method of the similarity measure of vectors, the trust model estimates the service recommendation credibility by measuring the behavior similarity between the service demander peers and service recommendation peers. Therefore, the model can identify dishonest feedback information effectively and has good time sensitivity for the peers' behavior. In addition, considering the sparsity of vectorial-similarity measurement, a precise scheme about the service recommendation credibility based on maximum likelihood estimation is proposed.Based on the deep analysis of current P2P topology optimization solutions, a research idea is proposed in chapter 4, which takes trust relationship as an important metric for topology formation. On this basis, an adaptive topology formation model based on profit sharing is put forward, which considers both service trust and routing trust while adjusting topology between peers. In the routing trust evaluation algorithm, according to the contribution of routing recommendation peers, the service demander peers share the transaction profit with them in order that the routing recommendation peers can attain the routing trust of their neighbors by routing search messages to them. This topology formation model not only provides an effective incentive mechanism to intermediate peers, but also improves P2P network which is in favor of good peers.To solve the problems, the poor scalability and providing the same level searching services to all the peers, existing in the traditional resource searching mechanism, the dissertation proposes a trust aware resource searching mechanism. In this searching mechanism, peers with unreliable resources are excluded in the searching stage which reduces bandwidth consumption by useless searching messages. Meanwhile, based on their trust value, the searching mechanism provides different peers with different level searching services to stimulate peers to provide reliable resource in P2P network.Finally, this dissertation studies the resource selection mechanism in P2P pervasive computing environment. In this computational market, service demander peers submit jobs to remote services for execution by paying the service provider peers. However, these resources may be unreliable inducing that there is a risk that submitted jobs may fail. So it proposes an opportunity cost approach for resource selection, which considers combination of explicit cost and failure penalty, models after the failure rate of resource by negative exponential function of trust value and provides a quantum method of opportunity cost. Comparing with Buyya's model, the approach effectively reduces job failure rate and saves total cost.Form the present problems in P2P application that the characteristics of network dynamicity, heterogeneity between peers and reliable resources, this dissertation deeply studies the autonomous and selfish features of peers, proposes a Tart Model which organically combines the market mechanism and trust mechanism, provides a trust aware resource transaction scheme for P2P applications, and shows a brand new view and fine perspective for solving the problem of P2P resource management.