Research On Quality Control Based On Game Theory In Crowd Sensing

At present,Internet of things devices are widely deployed in environmental monitoring,smart city and other scenes that need large-scale sensed data,which also provides strong support for the rapid development of crowd sensing.Due to the limited resources and computing power of a single device,the emergence of crowd sensing can greatly improve the acquisition of sensed data.While improving efficiency and reducing resource consumption,uneven data quality may occur in crowd sensing because of the heterogeneity of workers.Therefore,how to ensure the availability of data and strictly control the quality of data has become a hotspot that can not be ignored.To improve availability of data,quality control in crowd sensing has become the main research direction of this thesis.Research content of this thesis mainly includes two aspects,namely improving data authenticity and reducing data redundancy.The main research contents of this thesis are summarized as follows:1.In order to avoid low-quality data caused by untrusted workers,a quality control method based on multistrategy repeated game is proposed to improve the authenticity of data.Firstly,the quality control method proves the existence of sensing dilemma by modeling the interaction between workers and task requesters.Then,evolutionary game and Wright-Fisher model are combined,which is used to analyze the strategy evolution of both sides of game model.In addition,discount factor of repeated game and historical information of both sides are used to analyze behavior of wokers and task requesters.Finally,theoretical analysis and simulation experiments show that this method can more effectively motivate workers to provide high-quality data for a long time,and the evolutionary stability strategy is no longer a dilemma.2.Considering the limited task budget,a sensed data quality control method based on workers' coalition is proposed to reduce redundancy of data.Firstly,the method recruits workers based on trust degree of workers,and greedy algorithm is used to select the set of trusted workers under budget constraints.Then,workers' coalition partition in the trusted set is modeled as an coalition game.A stable coalition structure is found by using the coalition formation algorithm,and a reasonable methods of calculating payment are given based on Shapley value.Finally,simulated data and real data sets are used to verify the effectiveness of this method.