Research On Trust And Incentive Mechanism For The Uncooperative Behavior In Peer-to-Peer Network

The users enjoy great convenience due to the features of Peer-to-Peer such as autonomy, dynamic, robust and distribution. For distributed collaboration, resources sharing and parallel computing, Peer-to-Peer technology shows excellent performance, so it has become a very popular internet application mode. However, in Peer-to-Peer environment, the loosely coupled relationship between the nodes has also led to many security risks. The uncooperative nodes disregard the concept of "mutual cooperation, service shared, mutual benefit" that is advocated in P2P systems. Their bad behavior embodied such features as selfish, negative, malice. The root cause of these risks is the lack of supervision of the central server in P2P environment, resulting in the lack of trust relationship between nodes, so that the reliability of the resources and service cannot be guaranteed. To address these issues, there are effective ways to deal with by introducing the trust metric and deploying the incentive strategies. With the trust level, the nodes can choose a trusted node to interact, to avoid the propagation of illegal resource. The deployment of the incentive strategies improves the shared enthusiasm and contains the free riders. Therefore, there are important theoretical value and realistic prospect for P2P security system by study of the trust and incentive mechanism. At the same time, the research works are necessary for the issue of National Natural Science Foundation:"Research on P2P trust and reputation mechanism" and the project of National "863" High Technology Research and Development Program Foundation:" Research on key technology of P2P attack monitoring and warning with integration of Multi-feature"It is the leading clue of this dissertation for containing negative influence due to the uncooperative behavior in P2P environment. The starting points include the trust matric and incentive strategy. Modeling the trust relationship is studied. The combination for specific P2P technology and trust metrics is made. The method of urging share is discussed. The total content of this dissertation focus on the trust and incentive mechanism, and the following innovative results are obtained. (1) A trust model based on ant colony system is proposedFor existing unstructured global trust model, the way of obtaining the witness information often bring about redundant recommended path, and some long-chain of the trust, resulting in the inaccurate reputation values. To address this problem, the ant colony system is introduced. The pheromone and recommended reliability are associated. The inspired factor of the ant algorithm is viewed as the logic hops between the nodes. Find-path of witness peer is based on the ant algorithm. The trust paths between requesting peer and witness peer are optimized. The dishonest recommenders are punished by volatilizing their pheromone. With proposed method, the reputation metric is more accurate and the overhead of messages is decreased.(2) A optimistic routing strategy based on reputation is proposedThe traditional means to improve the routing efficiency in chord networks are optimizing the routing table. The routing decision-making is ignored. In existing improvement, the routing decision-making is one-side based on the greedy algorithm or the reputation metric. To this issue, a new routing method is proposed. To take into account the characteristics of routing in chord ring and the routing behavior can be described by Boolean logic, in proposed method, the routing behavior is evaluated with the FFP protocol and Beta reputation algorithm, the optimistic routing way is raised. By combining the greedy decision-making and reputation decision-making., the forwarding nodes are more rationally selected. The efficiency and reliability are reasonably compromised.(3) A reputation-sensitive topology adaption protocol is proposedThe exiting methods of constructing the unstructured topology in Gnutella-like P2P networks are not fully take into the characteristics of the nodes’behavior and the relevance of the resources. Cooperative nodes and non-cooperative nodes are in same status at topology reconstruction. For this issue, a new topology adaptive protocol is proposed, the trust metric is introduced. The neighbor nodes are selected based on their reputation. The possibility that uncooperative nodes become neighbor is greatly reduced. So the uncooperative nodes are gradually marginalized by the networks. Meanwhile, the-search process is enlightened by the similarity of the resources. The topology adaptive process as much as possible ensures better similarity of the resource between the neighbor nodes, thus the search is speeded. (4) An incentive mechanism based on the Optimazation allocation of resources is proposed.To address the issue of the unfair allocation of common resource due to the free-riders, a new incentive mechanism is proposed with a starting point of network bandwidth allocation. In proposed method, an optimal utility model is designed. The utility function is first decoupled. Then this function is solved with the method of the dual decomposition. To the difficulty of solving for the distributed environment, a new auxiliary variable and additional constraints are introduced, so the original problem is transformed into the dual problem. Each node can independently solve by the iteration. Compared with similar incentive mechanism, the proposed program not only ensures the nodes are motivated for share, but also achieve the system utility maximization.