Research On Location Privacy Protection In Mobile Internets

With the rapid development of GPS-enabled mobile devices and wireless communication technology,location-based service(LBS)has become one of the most popular applications in mobile internet.LBS is becoming an essential part of daily life.However,with location information,an adversary can easily infer users' lifestyles,such as living habit,health conditions,exercise habits,and hobbies,beyond just the locations.The potential abuse of location information is evolving into a serious concern.Recently,protecting the location privacy of mobile users has become a popular topic.However,with the development of LBS,new attacking means emerge in endlessly,location privacy protection still faces many challenges,including the location privacy metric,location privacy preserving technology,and the tradeoff between privacy preserving and quality of LBS services.Thus,in this paper,we address the location privacy issue in four different scenarios.The contributions of this paper are summarized as follows:First,although many approaches have been proposed to address the issue of location privacy protection,most of them relied on the trusted third party.We exploit the computation and storage capacities of the smart phones,and design a novel system architecture to provide a privacy-preserving LBS query.Based on the system architecture,we propose a novel metric called location indistinguishability to evaluate the privacy level of users in the proposed scheme.We then propose two dummy POI algorithms to generate a superset of the actual LBS query when the query cannot meet the privacy requirement.Evaluation results indicate that our methods achieve a high protection level with little communication cost increase.Second,a frequently proposed solution to protect location privacy suggests that LBS users need to exchange their pseudonyms over time in a protected region called mix-zone.Thus,we address the problem of optimal placement of multiple mix zones in large-scale city.We firstly propose a probabilistic model to quantify the privacy level of the LBS users.Then,we characterize properties and constraints of the optimization problem,and build a mix-integer programming model with the objective of minimizing the amount of time the users privacy level is lower than the privacy requirement.The optimization problem is NP-hard.Therefore,we propose a heuristic algorithm based on greedy strategy to strategically select mix zone locations.Simulation results based show that our solution yields satisfactory performance in reducing the risk of inferential attacks.Third,an adversary with big data analytical capacity can easily extract a great deal of side information from diverse mass data and then uses it to infer user's privacy.We thus propose a context-aware location information-sharing model.By means of data analysis,we discover three factors that can be used to infer a user's private location: geographic information,human mobility patterns and user preferences.Then,we propose a novel algorithm to integrate the above heterogeneous factors and provide a personalized check-in behavior prediction for each user.Based on the predictor model,we proposed a context-aware location-sharing framework that warns users when their actual sharing behaviors do not match their sharing rules.Fourth,with the development of smart phones,mobile users have shifted from being data consumers to being data providers,offering a new service model,i.e.,mobile crowdsourcing(MC).Different from LBS user,the sensing data that MC users provide must meet the quality requirement of the MC server,thus,the traditional loation privacy preserving tehnologies is not applicable for MC.Moreover,incentive mechanism is an important index to weight the performance of a MC system.The privacy issue and incentive meschanism must be considered simultaneously.To address the problem,we firstly introduce a double-sided combinatorial auction model,which only requires users' coarse locations,to assign MC tasks at a single point.Then,we find that the utility values of MC in different time slots are coupled.With the aim of maximizing the long-term system utility,we designe a context-aware participant recruitment mechanism,which offer the capability to dynamically adjust the participant recruitment mechanism depending on the ratio between the requesters and workers.Simulation results demonstrate that the proposed mechanism can achieve high utility while meet users' privacy requirement.