Private Map Matching: Realistic Private Route Cognition on Road Networks

The rush for personalized user information, triggered by the daily generation of a staggering amount of geospatial data from multitude platforms, is leading to an erosion of users' location privacy. To ensure the privacy of moving objects on road networks, most existing works do not enforce a strict constrain that the anonymized or perturbed geospatial points should lie on the road segments. Thus, rendering the results unrealistic. In addition, humans armed with GPS-enabled devices have proven to be effective sensors which can be beneficial to traffic monitoring and other crowd sourcing services. However, they are discouraged to participate due to privacy concerns. Based on these drawbacks, we make a case for fusing privacy to map matching in road networks. In this paper, we propose a novel privacy preserving map matching technique that utilizes hidden Markov model, tangent distance and geometric properties of road segments. Our technique harnesses location privacy by first performing map matching. Then based on a defined set of a user's sensitive locations, we introduce a new cost function and employ it to determine a minimum cost alternate private route in our shortest path problem. We demonstrate using the Microsoft Seattle real dataset the effectiveness of our technique and show that it provides realistic privacy in road networks.