A Clustering-Based Privacy-Preserving Method for Uncertain Trajectory Data

With the development of trajectory data mining, personal privacy information is facing a great threaten. To prevent privacy disclosure, some anonymous methods should be used before the raw trajectory data is published. Yet, most existing methods did not consider the uncertainty in trajectory, largely due to the inherent inaccuracy of data acquisition equipments, delayed update of mobile objects and so on. At present, only the (k, d) anonymity model solves the uncertain trajectory privacy-preserving problem. However, (k, d) anonymity model has two drawbacks. (1) It reckons the uncertain radius d as a constant value, which cannot capture dynamically changing uncertainty in trajectory data, (2) The object´s movements between two consecutive positions are simply modeled as linear, which is not accurate for many actual situations. In this paper, we propose a new method called Restore Its True to protect the privacy of trajectory data in publishing. It is the first paper to present this idea that transforming the trajectory to an uncertain area to cluster. First, we use the method in probability statistics to model the trajectory to an uncertain area. Second, we put the similar uncertain area into a cluster and sanitize them in an equivalence class. Finally, we compare the performance of our proposal with (k, d) - anonymity model in real datasets. The analysis and simulation results prove that our proposal can protect trajectories privacy effectively with lower information loss than (k, d) - anonymity model.