Research On Key Technologies Of Location Privacy Protection In Mobile Internet Services

With the deep integration of the Internet and mobile communication technologies,the mobile Internet has become an indispensable component of modern society.In the mobile Internet,users can use intelligent mobile devices(such as smartphones,tablet computers,vehicular on-board computers,etc.)to access the Internet in real time,and achieve the acquisition of information and services anytime and anywhere,thereby enjoying various convenient information services.The development of mobile Internet services makes the application available to users more and more abundant.Among them,users can not only enjoy the convenience brought by various services as information requesters,but also participate in information collection and data sensing as information providers.However,since users are usually required to submit their location information in mobile Internet services,they are facing serious threats of location privacy leakages,no matter whether users act as information requesters or information providers.It not only violates the personal privacy of users but also seriously hinders the popularization and development of mobile Internet.Therefore,it has become a key issue that needs to be solved urgently to ensure users’ location privacy in mobile Internet services.In this thesis,aiming at the differentiated participation roles of users in mobile Internet services,we analyze the privacy requirements and challenges for users participating in Location-Based Services(LBS)as information requesters and Mobile Crowdsensing(MCS)as information providers,respectively.Based on that,we conduct the research on key technologies of location privacy protection in mobile Internet services.Through the research work of this thesis,it can not only ensure the location privacy of users in using mobile Internet services so as to increase user trust in related services and encourage more users to participate in the services,but also promote the innovative development of mobile Internet services and expand their application fields.This is of great significance for encouraging Digital China Construction and serving the national cyberspace security strategy.The main contributions of this thesis are as follows:1.The existing location privacy protection methods for LBS ignore that the user’s location perturbation behavior may also carry private information,resulting in the location perturbation behavior being recognized by the adversary,which will cause the extensive sensitive information of the user,such as privacy preferences,personality,etc.to be revealed.To address this issue,we propose a privacy-enhancing scheme for LBS based on the improvement of Geo-Indistinguishability(Geo-Ind).Specifically,we first propose a new privacy definition,called Perturbation-Hidden,to provide a more strict privacy guarantee for locationbased services than Geo-Ind.Compared with Geo-Ind,Perturbation-Hidden is designed to eliminate the privacy leakage caused by the situation that the vehicular user’s perturbing behavior is recognized.Then,we design a privacy-preserving mechanism to implement the privacy definition,in which the plausible locations with user-specified attributes are taken as the candidate set,and the candidate set is randomly sampled to generate the perturbed location without revealing user privacy.Finally,the retrieval area is determined by dynamic programming to ensure the accuracy of LBS queries.Theoretical analysis proves that our mechanism satisfies the privacy definition of Perturbation-Hidden.Extensive experiments on simulations and a real-world dataset show that our proposal achieves 100% plausible pseudo-locations while ensuring high query precision and recall.2.The existing location privacy protection methods for LBS undermine the overall statistical location distribution of users after the location information protection,such that the service provider cannot provide distribution-related services to users.To address this issue,we propose a distribution-preserving location privacy protection scheme for LBS.Specifically,we first propose a new privacy definition suitable for LBS based on the inspiration of differential privacy,which is called Dist Preserv.It largely maintains the users’ overall location distributions on the basis of location privacy protection by requiring the reported locations and true locations to be indistinguishable in both Euclidean distance and distribution differences.Then we design a location privacy protection scheme for LBS,in which a location perturbation mechanism is designed to achieve Dist Preserv according to differential privacy exponential mechanism under the guidance of incentive compatibility.Finally,a dynamic programming method on the two-dimensional map plane is utilized to determine the retrieval area of LBS,thereby achieving high accuracy of queries with privacy guarantees.Theoretical analysis proves that the designed mechanism can achieve the definition of Dist Preserv and the property of incentive compatibility.Experimental explorations using a real-world dataset indicate that our proposal prominently improves the availability of users’ location distributions by over 90% while providing high precision and recall of queries.3.The existing privacy-preserving spatial distribution crowdsensing methods usually ignore the rational characteristics of users as participants in the system,resulting that users may not obtain satisfactory spatial distribution even if they provide true location information.To address this issue,we propose a privacy-preserving spatial distribution crowdsensing scheme based on the game theory.Specifically,we first model the privacy-preserving spatial distribution sensing in MCS as a game theoretic satisfaction form and define the satisfaction equilibrium for this game,which allows each user to make a trade-off between the level of privacy and the availability of the sensed distribution while considering the satisfactions of other users.Then we design two learning algorithms based on the implicit interactions among users through the platform to make users determine their strategies for the satisfaction equilibrium.The first LEFS algorithm is applicable to the case that users’ satisfaction expectations for the distribution are fixed.The second LSRE algorithm allows users to have dynamic satisfaction expectations for facilitating the equilibrium convergence and preserving more privacy for users.Theoretical analysis gives the convergence conditions and characteristics of the proposed algorithms.Experiments show the superiority and various performances of our proposal,which illustrates that our proposal can get more than 85%advantage in terms of the sensing distribution availability compared to the traditional spatial cloaking based solutions.4.The existing differential-privacy-based MCS task allocation methods ignore users’ preference for nearby tasks,which may not only lead to privacy leakages of users’ habitual locations during the long-term MCS participation,but may even lead to the failure of the MCS task allocation.To address this issue,we propose a privacy-preserving MCS task allocation scheme based on task dispersion.Specifically,we first propose a new privacy definition for the scenario of MCS task allocations,which is called Task-Dispersion.It allows users to select a relatively distant task for application with an adaptively adjusted probability,thereby avoiding the privacy leakage of habitual locations caused by the location concentration of tasks during long-term task allocation participation.Then we design a privacy-preserving mechanism for the user to achieve Task-Dispersion,in which the user determines the task to apply for through probability sampling according to his/her historical applications and the information of currently published tasks.Finally,based on users’ privacy-preserving submissions,the winner selection and payment determination methods are designed for the platform to complete task allocations with as few leftover tasks as possible.Theoretical analysis shows that our proposal can achieve the privacy level of Task-Dispersion and the properties of individual rationality and truthfulness.Experimental evaluations present the superiority and various performances of our proposal,which demonstrate that our proposal can reach up to 100% task completion rates with improved privacy protections.