Research On Incentive Mechanism For Mobile Crowdsensing Via Network Influence Diffusion

Mobile crowdsensing has emerged as an efficient paradigm for performing large-scale sensing tasks.Improving both quantity and quality of users is still a pivotal problem for mobile crowdsensing system.But crowdsensing needs to consume users' resources as well as the potential threat of privacy disclosure.Therefore,it is very important to design an effective incentive mechanism.First,the incentive mechanism based on social network influence diffusion in large-scale mobile crowdsensing applications is studied in this thesis.In order to solve the problem of the lack of users in mobile crowdsensing systems,the influence diffusion model in social networks is used to spread mobile crowdsensing tasks and recruit enough mobile users to complete mobile crowdsensing tasks.Therefore,this thesis proposes a general model of mobile crowdsensing system based on social network influence diffusion.This model mainly integrates reverse auction mechanism and influence diffusion model,and finds the approximate optimal solution of social cost minimization problem by approximate algorithm.Further,this thesis studies the impact of user data quality on the mobile crowdsensing on the same system model.This thesis designs a quality aware incentive mechanism based on network influence,which provides a comprehensive solution to improve the quantity and quality of participants simultaneously by designing social mobile crowdsensing architecture.An initial diffuser selection algorithm is proposed first to accommodate two wide-studied diffusion models,and then a lightweight,multi-metric comprehensive and parameter-free user quality evaluation method is presented.In order to take both social cost and user quality into consideration,a new criterion,called unit quality cost,is proposed.Finally,a reverse auction is presented to optimize the new criterion.This reverse auction takes both social cost and user quality into consideration.Through both rigorous theoretical analysis and extensive simulations,we demonstrate that the proposed incentive mechanisms achieve computational efficiency,individual rationality,truthfulness,and guaranteed approximation.