Semantics-aware Obfuscation for Location Privacy

The increasing availability of personal location data pushed by the widespread use of location-sensing technologies raises concerns with respect to the safeguard of location privacy. Toaddress such concerns location privacy-preserving techniques are being investigated. Animportant area of application for such techniques is represented by Location Based Services(LBS). Many privacy-preserving techniques designed for LBS are based on the idea offorwarding to the LBS provider obfuscated locations, namely position information at lowspatial resolution, in place of actual users’ positions. Obfuscation techniques are generallybased on the use of geometric methods. In this paper, we argue that such methods can lead tothe disclosure of sensitive location information and thus to privacy leaks. We thus propose anovel method which takes into account the semantic context in which users are located. Theoriginal contribution of the paper is the introduction of a comprehensive framework consistingof a semantic-aware obfuscation model, a novel algorithm for the generation of obfuscatedspaces for which we report results from an experimental evaluation and a reference architecture.