Research On Mobile Aggregated Data Publishing Based On Differential Privacy

With the rapid development of wireless communication technology and mobile location technology,users can obtain corresponding location services through mobile devices with GPS positioning function.Based on the application of location service,users gain great convenience by collecting the location information and sending relevant query results to users.However,once the sensitive information is analyzed by an untrusted third party,users' personal privacy will be inevitably exposed.At present,how to protect user location privacy in the process of data release has become a hot topic in the relevant research field.As a powerful privacy protection method,differential privacy technology guarantees the security and availability of data by adding a certain amount of noise to the data.This dissertation focuses on the differential privacy mechanism and researches the mobile aggregation data release solution.The main work is as follows:(1)To protect users' trajectory privacy in the process of mobile aggregate data publishing,this dissertation first proposes two basic schemes to protect location data publishing using differential privacy mechanism,namely direct noise perturbation and threshold noise perturbation.Then,by analyzing the data characteristics of mobile aggregate data,this paper proposes a hybrid noise adding scheme based on the static data release scenario,that is,different disturbance mechanisms are adopted in different time periods,so as to improve the utilization of privacy budget.For noise data,this paper designs a post-processing mechanism to improve the utility of the published data.In addition,it is proved that the whole process strictly satisfies the definition of differential privacy.Finally,the experimental analysis verifies that this scheme effectively protects the user's trajectory privacy.(2)This dissertation designs a scheme for protecting mobile aggregated data release using differential privacy based on dynamic real-time data release scenarios.First,based on the historical data of the past every day,the predicted value of the aggregate data to be released is obtained through the Prediction module.Then,using the differential privacy mechanism,an Adaptive Sampling module is designed to determine whether the original data is disturbed,and the appropriate privacy budget is allocated through the Privacy Budget Allocation module.In addition,in order to resist the influence of the noise injected by the Noise Adding module,according to the data characteristics of the moving aggregate data,the Grouping and Merging module is designed to merge the data in the adjacent region,so as to reduce the interference of the noise.Finally,the Post-processing module further improves the utility of the data.