Research On Privacy-Preserving Incentive Mechanism Based On Data Quality For Mobile Crowdsensing

In recent years,with the development of sensors and communication technology,mobile devices have powerful data collection and computing capabilities.In this context,mobile crowdsensing has attracted extensive attention.Unlike the traditional Internet of Things,which uses specially deployed sensors to collect data,mobile crowdsensing crowdsources sensing tasks to the participants holding the intelligent mobile device and utilizes the sensors in mobile devices to complete the sensing data collection.Compared with the traditional Internet of Things,mobile crowdsensing has the advantages of lower sensing cost,flexible sensing methods,larger sensing range,and richer sensing data types.However,the mobile crowdsensing system needs a sufficient number of users to provide reliable sensing data.Therefore,it is especially important to encourage users to actively participate in sensing tasks and reasonably evaluate the quality of user data.At the same time,the data uploaded by users will inevitably involve user data privacy,so the protection of user privacy should also be taken into account when designing a mobile crowdsensing system.Accordingly,this paper proposes a privacy-preserving incentive mechanism model.The main research and work are as follows:(1)The security truth discovery algorithm is studied.There may be a conflict of sensing data from different sensors,and the truth discovery algorithm can calculate the ground truth value and user data weight from the conflict data of various sources.The ground truth can be used as the result of the sensing task,and users’ weight can evaluate the users’ data quality.It can realize the core function of mobile crowdsensing.The security truth discovery algorithm proposed in this paper is realized by the double-masking method.It can calculate the result using the user’s encrypted data subject to protecting the user’s privacy.And it allows a certain number of users to exit.(2)Propose a method for evaluating the eligibility of sensing data.To prevent malicious users from uploading error data with large deviation,which will affect the accuracy of truth discovery results.This paper proposes to set up a data qualification evaluation before the truth discovery,which is designed based on secure multi-party computing.The method requires setting a qualified interval when publishing the perceptual task.The sensing platform can judge whether the user data is within the qualified range without knowing the specific sensing data of the user.To screen out qualified data in advance and improve the data quality of the data used by the truth discovery algorithm.(3)A privacy-preserving incentive mechanism model is designed.the incentive mechanism model is constructed based on data qualification evaluation and secure truth discovery.It realizes the overall process of task publishing,user recruitment,task execution,data qualification judgment,task result,data quality calculation,and user reward distribution.It can protect users’ privacy and make use of users’ data quality for differentiated reward distribution.And encourage high-quality users to actively participate in perception tasks.(4)A penalty mechanism algorithm based on data quality is proposed.To prevent the unqualified data with large deviation from occupying the computing overhead of the server,this paper designs a user punishment algorithm based on reputation.If the user’s reputation is lower than the warning threshold set by the system,the task reward of this round will not be published.If the reputation continues to fall below the elimination threshold,they are not allowed to participate in subsequent sensing tasks.