Research On Crowd Sensing Incentive Mechanism Based On Game Theory

Traditional wireless sensor networks collect data through fixed sensor nodes which are configured and deployed by the network owner.It has high network installation and maintenance costs,insufficient node coverage.With the widespread use of mobile devices such as smartphones,the concept of swarm intelligence has been proposed as a perceptual paradigm for mobile users with sensing and computing devices responsible for collecting and contributing data to various applications.Participants are voluntarily involved in Crowd Sensing(CS)and a sufficient number of participants can provide fine monitoring.How to encourage participants to join in Crowd Sensing and collect data is the core of Crowd Sensing,so we need to study the incentive mechanism.This thesis studies the incentive mechanism under the single-task scenario in crowd sensing and proposes a crowd sensing incentive mechanism based on regional coverage combining the reverse auction algorithm.In addition,this thesis studies the incentive mechanism under the multi-task scenario and proposes a crowd sensing incentive mechanism based on Coalition Game(CG)and a crowd sensing incentive mechanism based on Evolutionary Game(EG).The main research contents and innovations are as follows:(1)In order to improve regional coverage and user participation,this thesis proposes a regional intelligence-based incentive mechanism on account of reverse auction.In this mechanism,users participate in auctions according to participation cost.Under a fixed budget,the service platform selects the winning users in the auction and buys data from them according to the regional coverage maximization algorithm.Then,the users determine whether to continue to participate in the auction according to the rate of return,and the users who exit the auction calculate the expected rate of return and try to rejoin the auction.The simulation results show that the mechanism can increase the coverage area while ensuring the participation of users.(2)Under the multi-task scenario,an incentive mechanism based on coalition games is proposed.In this incentive mechanism,the users play games among the various task coalition and the users participating in the same sensing task belong to the same coalition.The process of game among users is also the process of selecting tasks.The users randomly select tasks to form the initial coalition structure,and the obtained remuneration will be reduced when the number of users participating in the same task is large.However,when the minimum requirements for remuneration of users are not satisfied,the users will re-select.The goal of the game is to maximize the total utility of the coalition structure.The simulation results show that the proposed method improves the total utility and obtains a stable task coalition structure.(3)Considering the cooperation between users,this thesis proposes crowd sensing incentive mechanism based on evolutionary game.The process of information interaction between participants in crowd sensing is modeled as an evolutionary game and the income of users is defined as the fitness in this game.Based on the rules of "survival of the fittest" in evolutionary games,the users with higher returns are constantly being generated and the users with low returns are constantly being eliminated,which will encourage users to interact with information so that the total benefit of users will be maximized and the evolutionary game will eventually reach equilibrium.The simulation results show that the mechanism can improve the total user revenue and the system stability under the constraints of the budget threshold.