| Mobile crowdsensing has become an efficient paradigm for performing lage scale sensing tasks with low cost,and it has attracted close attention from academic and industrial circles at home and abroad in recent years.Mobile crowdsensing depends on the participantion of smart divice users.Users collect the data through various sensors embedded in the smart devices,and submit the data to the platform.In order to complete the specified sensing task,the users need to pay their own time and energy and consume the energy of the device.Therefore,it’s very important to design effective mechanism to compensate users.This thesis focuses on incentive mechanism design for mobile crowdsensing based on reputation system and social network.First,as the ability of users changes over time,the quality of sensing data is unstable,which affects the quality of task completion.In order to solve this problem,the thesis proposes a fine grained reputation-based crowdsensing framework,which can obtain users’ long term ability reputation by assessing data quality after they finish tasks.The crowdsensing incentive mechanism based on fine-grained ability reputation system is proposed based on this novel mobile crowdsensing framework.The users’ long term ability reputation is modeled by reputation system.Then the reverse auction is used to select the winners to perform the tasks.Through both rigorous theoretical analysis and extensive simulation experiments,the proposed mechanisms are proved to achieve properties of computation efficiency,truthfulness,individual rationality,whitewashing proof and low approximation ratio,and the social cost of FAR-Beta is 79.16% and 77.82% of social cost of Cost Min and Ability Max,and the average ability achievement ratio is 146.32% and 217.16%of those of Cost Min and Ability Max on average respectively,and the social cost of FAR-Peer is80.67% and 79.30% of social cost of Cost Min and Ability Max,and the average ability achievement ratio of FAR-Peer is 150.89% and 223.93% of those of Cost Min and Ability Max on average,respectively.Moreover,traditional crowdsensing platform often fails to comlete a large number of tasks due to the insufficient of users.In order to solve this problem.this thesis proposes a two-tiered social crowdsensing framework which can recruit social neighbors of the selected registered users through diffusing the sensing tasks to their social networks.The two tiered mobile crowdsensing incentive mechanisms are proposed for offline and semi-online scenario respectively,which select agents to diffuse tasks into their social networks to recruit their social neighbors.Through both rigorous theoretical analysis and extensive simulation experiments,the incentive mechanism under offline model is proved to achieve properties of computation efficiency,truthfulness,individual rationality and low approximation ratio,and can obtain averagely 83.4% value of approximate optimal untruthful offline algorithm,and the incentive mechanism under semi-omline model is proved to achieve properties of computation efficiency,truthfulness and individual rationality,and can obtain averagely 51.1% value of approximate optimal untruthful offline algorithm. |
