Research On User Location Privacy Preserving Technique For Location-based Service

With the rapid progress of positioning and wireless communication techniques as well as the development of smartphones, nowadays people can enjoy the benefits from various tailored and personalized LBS (location-based service) anytime and anywhere. A typical example of LBS is the answer to a location-related question such as "where is the nearest gas station?". While enjoying various LBS, users also face the threats of location privacy disclosure. This is because some malicious adversaries may gather a user’s exact location illegally by eavesdropping the user’s query to an LBS provider (LSP) so as to disclose his/her privacy information. Various solutions have been proposed to enable mobile users to access LBS whilst preserving their location privacy. Most of these solutions can be classified into adopting one of the two main architectures, i.e., trusted third party based architecture and mobile P2P (peer-to-peer) architecture. The bottleneck of the former is the third party in terms of computing and communicating resource when networks grow large, whilst the later has to endure either low anonymization success rates or expensive communication overheads and also suffers from kinds of attacks.In this dissertation, we report our investigation on location privacy preserving techniques within the mobile P2P architecture. In order to enhance the protection of location privacy while still deliver good quality of location based service, we try to improve the existing solutions in terms of (1) deploying a novel strategy when searching for candidate locations;(2) reducing P2P communication between mobile users;(3) allowing different trust assumptions among users; and (4) considering the semantic location affection. In summary, we make the following contributions.(1) In order to address the low anonymizing success rate and long waiting time when generating the cloaked region, we have proposed a dual-active location privacy preserving algorithm, in short as DA algorithm. It allows users to actively send their own location information and their maintained location information to other users. Moreover, it enables users to control the location’s valid period by utilizing several user-defined parameters. Compared with some existing algorithms, our DA approach uses the least anonymizing time and has the best anonymization success rate.(2) In order to enhance our DA algorithm in terms of communication overhead and the anonymization quality of the generated CR (cloaked region), we propose an enhanced dual-active algorithm, in short as EDA algorithm. The EDA algorithm deploy three novel strategies, namely location negotiation, location prediction and LRHP (latest-record-highest-priority) broadcasting, so that our EDA algorithm can improve the performance in terms of anonymizing time with minimized communication overhead.(3) Since most of existing solutions are based on the trust assumption that all mobile users are trustworthy which is not practical in the real world, we present x-region as a solution to preserve the location privacy in a mobile P2P environment where no trust relationships are assumed amongst mobile users. The main idea is to allow users to share a blurred region known as x-region instead of their exact locations so as to defend the impersonation attack. We also develop three algorithms for generating an x-region, namely, benchmark algorithm, weighted expanding algorithm, and aggressive weighted expanding algorithm.(4) In order to address the semantic location attack, we propose SALS, a semantics aware location sharing framework which is built on our aforementioned x-region solution. Different from the previous cloaking techniques, SALS considers the semantic location which can influence the distribution probability of a user’s locations and try to generate a CR in which the distribution probabilities are similar with each other so as to defend the semantic location attack.