Research On Free-riding And Incentive Mechanism In P2P Network

As a new computational mode, P2P (peer-to-peer) network modifies the traditional C/S mode and is used more and more widely. Nodes can join in and quit network at any time and share resources voluntarily.But it causes that there are many free-rider and non-cooperative nodes. Tragedy of Common is very common in the network, and there still exist a lot of unreliable services and malicious actions, which seriously affects the performance of P2P network. Therefore, to improve the performance of P2P, we must analyze the nodes'action and encourage the cooperation among nodes. And how to design an effective incentive mechanism to restrain free-riding and encourage the cooperation is one of the hottest issues of P2P research in recent years.On the basis of summary and comparison of the existing incentive mechanisms, and to overcome the shortcuts of current mechanisms, this paper proposes two mechanisms based on immune theory and neutral network theory. It aims at providing discriminatory service according to the contribution of nodes, making the maximum utility of the P2P network.The mechanism based on immune theory uses biological knowledge and model to solve the P2P networks. It classifies and supervises the nodes according to their type and function. When a node applies for resource, its neighbor nodes calculate its contribution, then establish the immune model, and calculate the affinity to invoke immune response process, restrain free-riding nodes and promote cooperation between nodes. When measuring the contribution of the node, considering the size, number and download count of shared resources, as well as forwarding and hit- on messages neural network. The experiments show that it should be used in low-cost and low network cost P2P network systems, which restrain the free-riding nodes not very strictly.The mechanism based on neutral network theory mainly focuses on the judge of the nodes'contribution. Establishing the Neural network model and training the sample to self-learn and calculate the results. It can evaluate the contribution of the node rationally and correctly. The experiments show that it should be used in complex P2P network system, which restrain the free-riding nodes strictly.The two incentive mechanism can overcome server-bottle, white wash and collusion. The experiments show that they can effectively restrain the free-riding nodes and encourage the cooperation.