A Privacy-Preserving Framework for Mining Continuous Sequences in Trajectory Systems

Abstract

The popularization of mobile devices and satellite navigation systems has brought new opportunities and challenges to many location-based services (LBSs), intelligent transport systems (ITSs) and fleet management systems. Those systems store trajectory data in moving object databases and trajectory data warehouses. Trajectory datasets can be used in many data mining tasks, e.g., for mining traffic patterns, frequently visited places, and so on. However, such data mining tasks introduce security and location privacy threats to companies (data owners) and mobile objects that generate trajectories. In this paper, in order to overcome the location privacy threats, we present a technique that blurs trajectories according to user-defined privacy profiles and a data mining algorithm called MCSPP that is capable of mining continuous sequences from blurred trajectories. To test that our approach works correctly, we have implemented an experimental environment. Moreover, the conducted experiments have shown a good performance and scalability of the MCSPP algorithm.