Research On Location Privacy Protection Technology For Sporadic Query

In recent years,with the popularization of smart devices,mobile applications have been developed rapidly.These applications rely on location-based services to provide personalized service contents by using users' coordinate or semantic/logical location information.The release of users' location information has raised serious privacy issues.Researchers have conducted extensive research in defining the location privacy and constructing optimal tradeoff between privacy and functionality.Existing definitions of privacy are mainly based on information theory or on differential privacy,and most of the researches are based on the coordinate locations of users.However,these definations cannot measure privacy accurately and effectively,even in situations that only the users' coordinate locations are considered.Besides,although the researchers introduce metrics for privacy and quality of service for applications based on semantic/logical locations,these metrics are problematic due to some reasons.This thesis deals with the problems above in the context of sporadic queries.For scenarios where only users' coordinate locations are considered,this thesis makes analysis on differential privacy,and believes it is too strict and has limitations such as long-tail problem,thus it is cannot be used directly for location privacy.This thesis proposes an improved defination based on differential privacy,constructs the optimal location privacy protection mechanism(LPPM)and studies its solution based on a zero-sum game between users and the attacker.Finally,the proposed optimal LPPM is compared with existing optimal LPPMs based on different definitions,so as to verify the correctness of the proposed privacy definition and the effectiveness of the proposed optimal LPPM.This thesis discusses the limitations in the metrics of privacy and quality,by proposing the concepts of granularity and granularity independence(GI),and discusses a sufficient condition to achieve GI.This thesis proposes an optimal LPPM that meets GI,and also proposes an acceleration mechanism.Both theories and experiments are given to verify the mechanisms above.Finally,this thesis makes a statistical classification of the existing location-based applications using NLP,and verifies the significance of the research on privacy preserving on semantic/logical location scenarios.Finally,based on the theoretical research results,this thesis makes a detailed analysis on user's location information in the Android system,and implements a prototype system which follows the Client/Server model.The client side mainly is composed of a location information interception module and a user interaction module,and the server side is composed of a privacy protection module.The two sides together provide real-time location privacy protection.Finally,based on existing mobile applications,functional and performance tests are carried out to verify the effectiveness and the feasibility of the prototype system in real-world deployment.To sum up,this thesis targets at the limitations of privacy mertic in coordinate location scenarios,and the limitations in metrics of privacy and quality in logical/semantic scenarios.To address these limitations,this thesis proposes a privacy definition based on differential privacy,and introduces the concept of GI for optimal LPPMs.Finally,a prototype location privacy protection system is implemented within the Android system based on Xposed.