Research On Location Semantics Based Privacy-preserving Technology For Location-based Services

With the widespread use of mobile devices(e.g. smart phone) and rapid development of positioning、wireless communication technologies, Location-based Services(LBSs) have become more and more popular. Since integrating with location information, LBSs can provide users with various services, which are close to their daily lives and satisfy their personalized privacy requirements. However, the disclosure of users’ exact locations when requesting service, and further the disclosure of their behavioral patterns、hobbies and health conditions have raised users’ concern about their personal privacy. How to guarantee users’ privacy while providing LBS becomes a serious problem.Most existing privacy preserving schemes didn’t consider location semantics when performing the privacy preserving process. As a result, they can’t resist semantics based attacks, which makes users still under threaten of “sensitive information disclosure”. Therefore, this thesis designs various methods to characterize location semantics and measure semantic security, and presents semantics-aware location privacy preserving mechanisms. At the same time, our mechanisms can provide personalized privacy protection for users according to their privacy requirements. The main contributions are as following:(1) The definition of location semantics and characterizing methods for it are proposedExisting semantics-aware location privacy preserving schemes haven’t describe location semantics with a uniform standard. Therefore, we propose the definition of location semantics and design effective methods to characterize it.(2) A semantics-aware location privacy preserving mechanism based on distance of vectors is proposedThe mechanism picks out a feature vector to characterize location semantics by using a feature selection algorithm and uses cosine value to measure the semantic dissimilarity of any two locations. To resist semantics based attacks, this mechanism cloaks a user’s exact location with μ-1 other locations, which have certain semantic dissimilarity with it. At the same time, we may generate dummy queries to satisfy the k-anonymity principle during the privacy preserving process if necessary.(3) A semantics-aware location privacy preserving mechanism based on information entropy is proposedThe scheme adopts a tagging method to characterize location semantics, based on which, we design a semantic sensitivity measuring algorithm based on information entropy to evaluate a region’s semantic security. Considering the efficiency of privacy preserving process, this mechanism partitions the whole map into hierarchical regions and computes the semantic sensitivity of each region in advance. Then, during a cloaking process, we just need to traversal the regions covering users’ exact locations from small size to large size and pick out the one which is semantic secure and satisfies their privacy profiles as their cloaked regions.(4) Evaluating the semantics-aware location privacy preserving schemesWe perform theoretical security analysis for the proposed semantics-aware location privacy preserving schemes, and conduct experiments for them to evaluate their effectiveness and performance.