Research On Incentive Mechanism Based On Potential Game Theory In New Network Applications

Game theory is an important theoretical tool to design traditional network application incentive mechanism.However,game theory usually focus on two extreme situations when designing traditional network application incentive mechanism.One is assuming that all users in the network are selfless,and maximize overall network interests.The other is assuming that all users in the network are selfish,and maximize own individual interests.However,new network applications such as mobile crowd sensing and vehicle ad-hoc network(VANET)are coupling physical relationships and social relationships between nodes.Device nodes form a physical network,while owner nodes form a social network.Users in this new network applications have rich social relationships,and are no longer completely selfless or completely selfish.Focus on this problem,social group utility maximization(SGUM)model is a new application framework for game theory and integrated with social network.This framework can define users' social group utility function,and maximize users' social group utility,in order to solve the problem between network utility maximization(NUM)and non-cooperative game(NCG).This paper considers two scenarios in the new network applications.One is the problem of user's strategy selection in mobile crowd sensing.It is necessary to incent user nodes participating in sensing task to improve network utility.The other is the problem of vehicle's selection of pseudonym change in the VANET.It is necessary to incent more vehicles to participate in pseudonym change process.Based on potential game theory and SGUM model,this paper studies and designs incentive mechanism in mobile crowd sensing or VANET pseudonym change to stimulate users' enthusiasm and improve system utility.This paper's main work as follow:1)Constructing a social group incentive mechanism by mining the physical relationships and social relationships between nodes in network.Defining social group utility function for each node,with which node focus on not only its own utility,but also the utility of the users have social relations with it.2)By establishing physical relationship network and social relationship network between nodes,SGUM game is constructed to maximize each node's social group utility function.SGUM game transforms into potential game when constructing potential function for it,thus is can achieve a pure social aware Nash equilibrium in incentive mechanism.This SGUM game is applied to two specific problems,one is the incentive problem in the mobile crowd sensing,and the other is the incentive problem in pseudonym change of VANET.3)This paper designs a distributed incentive algorithm based on Markov chain to promote a stable social aware Nash equilibrium among users.The algorithm is a heuristic algorithm that iterative updates node's strategy to optimize node's social group utility continuously.And the social aware Nash equilibrium obtained by the algorithm is a strategic combination close to optimal solution.4)This paper builds the experimental environment of mobile crowd sensing and VANET pseudonym change.Then compares the performance between SGUM,NUM and NCC.Experimental results show that the social group utility of the incentive algorithm have been significantly improved after bringing in social relationships.