Font Size: a A A

Game-theoretic Analysis Based Trajectory Privacy Preserving In Location Services

Posted on:2017-03-04Degree:DoctorType:Dissertation
Country:ChinaCandidate:Y H HeFull Text:PDF
GTID:1108330488457180Subject:Computer system architecture
Abstract/Summary:PDF Full Text Request
Location services have gradually penetrated our daily life, and provide with abundant appli-cations, such as location aware sensing for natural environment, public infrastructure, and social activities, location-based query services for taxi, navigation, ad pushing, and trajec-tory publishing for analysis and mining by researchers and companies. Location services bring convenience to our daily life, but they disclose the location information of mobile users. Specifically, location service providers can obtain a user’s locations or trajectories di-rectly or indirectly in the process of collecting sensing data, providing location based query services, and trajectory publishing. In addition, location privacy can be also leaked by the inference attacks and the interactions between users.This thesis focuses on privacy issues in three types of location service applications. We first summarize and analyze existing location and trajectory privacy protection approaches. Then based on the analysis, we use game theory to design appropriate trajectory privacy preserving mechanisms for the three types of location service applications. Our research interests consist of three aspects:(1) For location aware sensing services, our study focuses on designing a trajectory privacy preserving mechanism for crowdsensing location services. In crowdsensing location ser-vices, sensing accuracy and user privacy are contradictory:the more information about natu-ral environment, public infrastructure, or social activities that collected by service providers, the more accuracy the obtained sensing information is; while the more abundant sensing information a user has uploaded, the more likely to reveal the user’s location privacy. To address this challenge, this thesis proposes a privacy preserving location aware data upload mechanism, which considers balancing the service quality of crowdsensing and the location privacy of users. In this mechanism, an incomplete information game model is used to study the user upload behavior game, where each player considers a tradeoff between the crowd-sensing service quality and its own location privacy to decide whether or not to upload. Based on Nash equilibriums in this game, we analyze the relationships between a user’s up-load behavior and the crowdsensing service quality, and between the user’s upload behavior and its location privacy leakage. Based on the analysis, we design a suitable upload strategy for each user, which can not only meet the basic requirement of the crowdsensing service quality, but also provide a personalized privacy preservation. Through extensive simulation study based on real world trajectory data, we show that our mechanism can reconcile crowd-sensing service quality with location privacy, instruct each user to select the optimal upload strategy, and maximize each user’s utility.(2) For location-based services(LBS), we study how to protect trajectory privacy in loca-tion based query services. Existing location privacy protecting techniques either rely on a trusted three party to anonymize or cloak a user’s query request, or allow users gener-ate dummy queries to protect location privacy and query privacy. For the former case, the trusted party may sell users’ private information to others to earn extra rewards, and a user may be unwilling to trust the third party. For the later case, dummy queries will increase a user’s storage, communication and computational overhead, and they can be eliminated by some network tracing techniques. Therefore, this thesis proposes a privacy preserving query method based on community effort, in which a user cooperates with nearby users to form a group, and group members jointly develop a query generation strategy for protecting loca-tion and query privacy. In our mechanism, a Bayesian game model is used to analyze the effect of different query strategies on k-anonymity success ratio. By setting reasonable game parameters, this mechanism can motivate users to collaborate, improve the k-anonymous success rate, and maximize the utility of users. From Nash equilibrium analysis and simu-lation experiments, we show that our mechanism can instruct users to generate an optimal query strategy to guarantee the k-anonymity success rate for both the cases of the number of group members larger and equal to k and less than k.(3) For location publishing services, we propose a trajectory privacy preserving mechanis-m in trajectory big-data publishing services. Although many trajectory privacy preserving approaches has been proposed, such as adding dummy trajectories, reducing GPS samples in trajectories, or adding noise to GPS samples, these approaches focus on some specific trajectory analysis scenarios. Trajectory analysis scenarios have different requirements on the truthness of trajectory data, which also cause different effectiveness of privacy protecting techniques on trajectory analysis scenarios. In addition, different attack strategies will result in quite different performance under a privacy preserving strategy. In order to address these challenges, we design a privacy preserving strategy selection algorithm to instruct a third party data center to select the optimal denfense strategies. We first use complete and incom-plete information game models to capture the behavior of the adversary and the defender. Through the attack and defense game, we analyze the effectiveness of privacy preserving s-trategies under different attack strategies. Then based on the analysis results, an algorithm is proposed to instruct the third party data center to select the optimal defense strategies under different data truthness requirements, maximizing the utility of the defender. Our experi-mental results indicate that our algorithm can obtain a higher defender’s utility compared to other approaches.
Keywords/Search Tags:Location services, location privacy, quality of service, game theory, Nash equi- libriums
PDF Full Text Request
Related items