Personalized Privacy Protection For Mobile Crowdsensing Systems

With the popularization of smart devices,the mobile crowdsensing technology has been widely used in various fields and becomes a popular research issue in recent years.Mobile crowdsensing can allocate tasks and collect data by leveraging each mobile user as a basic sensing unit.However,mobile users may face the risk of privacy leakage when participating in mobile crowdsensing systems(MCS).The participation of mobile users will be degraded seriously if MCS ignore the privacy problems.Hence,many researchers have focused on the privacy protection problems for mobile crowdsensing systems and some meaningful works have been proposed.However,the existing works assumed that all users adopt the same privacy protection level for their sensitive data,which is beneficial to the platform management but ignoring the diversity of privacy requirements from different users.Using the same level of privacy protection will only make some users' privacy requirements unsatisfied,while other users' data will be overprotected.Therefore,we start from the diversity of user privacy and implement the research on personalized privacy protection for mobile crowdsensing systems.We aim to solve the task allocation,user incentive and data aggregation problems involved in MCS while satisfying the personalized privacy requirements of users,and finally establish a full-cycle privacy protection theory that supports the task allocation?user incentive and data aggregation.In order to achieve personalized privacy protection,we adopt the differential privacy technology to perturb the sensitive data of users such as location information and sensing data(heart rate)involved in MCS.Each mobile user can select the privacy budget according to his privacy requirement and perturb sensitive data locally before uploading it to the platform.(1)Aiming at the task allocation problem,we propose a Probabilistic Winner Selection Mechanism(PWSM)to minimize the total travel distance with the obfuscated information from workers,by allocating each task to the worker who has the largest probability of being closest to it so that the task performance efficiency can be improved.(2)Aiming at the user incentive problem,we propose a Vickrey Payment Determination Mechanism(VPDM)to determine the appropriate payment to each winner by considering its movement cost and privacy level,which satisfies the truthfulness,profitability and individual rationality.(3)Aiming at the data aggregation problem,we propose a History based Expectation Maximization Mechanism(HisEM),which integrates the user's historical data into truth discovery so that the data aggregation robustness can be improved and achieves accurate estimation for truths of tasks.Finally,privacy analysis and extensive experiments on the real-world datasets demonstrate the privacy and effectiveness of the proposed mechanisms.