Research On Evolutionary Game Model For Trusted Cooperation In Opportunistic Network

Opportunistic network is a new form of distributed Ad-Hoc network.It has features of distributed and Ad-Hoc,which is benefit to build network rapidly.Currently,Opportunistic network has been widely applied to many communication scenarios like deep space communication and in-vehicle network.In recent years,with the popularization of mobile terminals,some new-type social networks based on opportunistic network occur one after another.These show that opportunistic network has a huge foreground in Ad-Hoc network field.However,the communication of opportunistic network depends on nodes cooperating with each other.In opening network,the process of message delivery would be disturbed by selfish or malicious nodes,which may badly disorder network transmission.Therefore,identify all abnormal nodes from network accurately and promote abnormal nodes to cooperate,which is an urgent problem.In order to curb the influence of abnormal nodes on the opportunistic network,this paper proposes a trust evaluation model with multi-dimensional convergence.Trust evaluation model includes nodes behavior monitoring mechanism and trust degree quantification method.Monitoring mechanism is performed among nodes to detect and record interactive data.Trust degree quantification method is used to quantify nodes' forwarding capability and credible degree from different dimensionality like connectivity,health condition and quality of service.At last,the model uses different dimensions of trust to optimize routing policy of nodes.For instance,messages will be sent to nodes with high transmit capacity and high credible degree.And abnormal nodes will be isolated from network.For the convenience of determining abnormal nodes,we fusion these three dimensions of trust into a synthetic reputation.As time goes on,those isolated nodes will involve in collaboration actively to acquire forwarding service.To research the evolutionary problems of trust cooperation between nodes,evolutionary game tools are introduced.We construct evolutionary game model to promote all nodes to involve in collaboration actively by self-evolutionary.Single phase game analysis shows that the process of message forwarding is a classical prisoner dilemma.In order to guide the evolutionary game developing towards positive,we merge reputation incentives into benefits matrix and policy update mechanism.Specifically,reward and punishment factors are introduced into the game proceeds function to reward cooperation behavior and punish noncooperation behavior.In policy update mechanism,whether nodes change their policy or not based on their previous gains.If it is changed,they need to select a role model.The process of filtering role model also references node's reputation.In this paper,we use The ONE which is an open source simulation toolkit for opportunistic network to verify the effectiveness of the proposed model.We deploy the model to several classical routing algorithms and design different simulation environment to simulate the experiment.Simulation results show that trust evaluation model can identify abnormal nodes accurately which provides positive motivation for evolutionary game model.Evolutionary model for trust cooperation has good scalability and works well in different network environment.