Location Depersonalization for Location Privacy Protection in Location-Based Services

A spatial region with K different footprints indicates that it has been visited by K different people. An adversary may be able to identify all these users, but will not know who was there at what time. This observation gives us a new direction to investigate location depersonalization in location-based services (LBSs). We propose to leverage users’ historical location samples, each being a footprint, to cloak their current location. For each location/trajectory reported for LBSs, we ensure that it has been visited earlier by at least K − 1 other users. In contrast to existing techniques which depersonalize a user’s location based on her current neighbors, our approach is able to provide a certain level of guarantee that a user’s location information, either a single location sample or a time-series sequence of them, cannot be correlated with restricted spaces such as home and office to derive who’s where at what time. In addition to location privacy protection, using footprints for cloaking can significantly improve cloaking resolution and allow mobile nodes to report their location only when they are engaged in LBSs. We present novel algorithms for single location sample cloaking and trajectory cloaking, and evaluate their performance under various conditions using location data generated based on real road maps. Our results show that the proposed techniques have a minimal impact on the quality of LBSs.