The Research On Restraining Free Riding In Peer-to-Peer Networks

With the advancement of information technology and speedy prevalent of Internet, Peer-to-Peer Networks have increasingly large scales and more and more ranges of applications. In a P2P system, each user node has dual identity of being a service provider and a client, and the efficient running of the system depends on contribution of every node, in other words, providing service for other nodes. However, there are some nodes that only get services provided by other nodes without contributing local resources for the system, which lead to free riding phenomenon. Studies have demonstrated that the existence of numerous free riding actions has significant effect on robustness, scalability and usability of the networks.For the purpose of restraining the free ridings of nodes in P2P networks, many domestic and overseas researchers have proposed various effective mechanisms to constrain the free ridings of nodes and contribute local resources to the systems in order to boost the mutual cooperation between nodes and healthy development of networks. Currently, the main restraining free riding approaches can be divided into four classes:based on virtual currency, based on reciprocity, based on reputation and based on game theory. Moreover, game theory mechanism is the most research hot spot in recent years, and also is the theoretical basis of the restraining method proposed here.In this paper, we refer to the reputation-based mechanisms to restraining free-riding behaviors and designs a novel graded punishment model. In the model, a node with higher reputation grade will suffer smaller penalty than with lower reputation grade when takes one free riding conduct, and nodes in the P2P system consequently prefer contributing for better reputation grades for their own benefits. Then the game theory is introduced to analyze the free riding problem, and nodes' behaviors in P2P networks are modeled as an infinite repeated game. Subsequently, game theoretical analysis proves the existence of Nash Equilibrium solutions of rational nodes in a system with diversified nodes. In addition, when some requirements are met, rational nodes have much more incomes taking contributing strategy than taking free riding, so they are motivated to cooperation and sharing. At last, we use OverSim simulator to implement the system with graded punishment mechanism, and the experimental results shows that the mechanism is effective in restraining free-riding behaviors and improving cooperation between nodes in a P2P system with simultaneous existence of enthusiastic nodes, selfish nodes and rational nodes.