Differentially private publication scheme for trajectory data

Trajectory data. like human mobility trace, in participatory sensing is of vital importance to many applications, like traffic monitoring, urban planning and social relationship mining. However, improper release of trajectory data can incur great threats to user's privacy. Recent researches have adopted Laplace mechanism to achieve differential privacy which can guarantee that small change of one record in database will not breach a user's privacy. However, existing work cannot guarantee privacy perfectly because a randomly picked noise will not contribute to a meaningful trajectory data release and people need to hide their visits to certain sensitive area. In this paper, we propose a differentially private trajectory data publishing algorithm aiming to protect the privacy of sensitive areas. Privacy analysis show that the proposed scheme achieves differential privacy and experiments with real trajectory data exhibits that the proposed scheme achieves good data utility and is scalable to large trajectory databases.