A Study Of Location Privacy Protection In Location-based Services

In recent years, due to the diversity of presentation forms and convenience of user experience, Location-Based Services(LBSs) have attracted increasing attentions. The LBS server can provide plenty of valuable service data for users based on their submitted location information, significantly facilitate users’ daily life and work. However, the upload of location causes a series of privacy issues. Through the collection of historical locations, users’ sensitive information, such as daily activity, health condition and social status, can be mined and exposed. With the reinforcement of users’ privacy consciousness, people hold a reserve attitude towards LBS because their location information may be misused or abused, which heavily hinders the development of LBS. Therefore, how to protect users’ privacy has become a world-wide hot topic. In this thesis, we have a deep research on the location privacy protection in LBS. The main contents are listed as follows:1. The thesis has an explore on the main problems and techniques in location privacy protection. We study the existing schemes and give a brief introduction of the common mechanisms, techniques and schemes. In addition, we research on the design objective, system architecture and threat model. We focus on location anonymization and give the different implement structures. Besides, we analyze the advantages and disadvantages of each structure based on the possible attacks.2. A collaborative system based on cache and location anonymization has been proposed. To improve users’ location privacy, we introduce the conception of cache. Users compose a collaborative group and communicate with each other through a mobile ad hoc network. In this group, they cache their used service data instead of discard them and share the cached data with others. In this way, many users’ requirements can be satisfied locally in the group before sending query message to the LBS server. Thus, these users don’t need to worry their privacy be revealed by LBS server. Moreover, to improve the utilization of cached data, a threshold-based module is adopted. When and only when the requirements cannot be met locally, users have to access the LBS server.For users of this type, we propose a Dummy Selection Algorithm(DSA) based on the distribution of cached data. DSA selects the locations in which data haven’t been cached. The employment of dummy locations enhance the anonymity of users’ real locations and increase the cached data in the system. Furthermore, to resist some targeted attacks from the LBS server and increase the cached data, we also propose a enhanced DSA, which determines dummy locations based on the query overlapping ratio of each location. Specially, only if the query overlapping ratio of a location is lower than the threshold pre-defined by users, this location can be a dummy candidate. At last, we test the performance of the cache strategy and DSAs through a simulation conducted by C++ language.3. A privacy-preserving system based on dynamic pseudo-IDs in LBS has been proposed. We indicate that long-term pseudonyms fail to protect users’ privacy. To address the problems caused by long-term pseudonyms and prevent the inference attacks aided by long-term observation and side information, in this system, users change their pseudo-IDs in query messages to break the link between the real identity and location, achieving identity anonymization. Specially, users obtain pseudo-IDs from the Trusted Authority(TA) and then pre-stored them for the application of LBS in a period. TA generates each pseudo-ID through the XOR operation of two hash chains. For application and management, each pseudo-ID is attached with a corresponding certificate and each certificate is valid in only one time slot. By this means, LBS server can verify a user’s validity anonymously to avoid responding to invalid users. Besides, for the malicious users, LBS server can report their pseudo-IDs and related certificates to TA and revoke the malicious users with TA. At last, we conduct a simulation experiment under the Levy mobility model and side information aided inference attack model, demonstrating the effectiveness of this scheme; we also test the cryptographic operations in a real smart phone, showing that the proposed scheme is feasible to implement into mobile devices.