A Clustering-Based Approach for Personalized Privacy Preserving Publication of Moving Object Trajectory Data

With the growing prevalence of location-aware devices, the amount of trajectories generated by moving objects has been dramatically increased, resulting in various novel data mining applications. Since trajectories may contain sensitive information about their moving objects, so they ought to be anonymized before making them accessible to the public. Many existing approaches for trajectory anonymization consider the same privacy level for all moving objects, whereas different moving objects may have different privacy requirements. In this paper, we propose a novel greedy clustering-based approach for anonymizing trajectory data in which the privacy requirements of moving objects are not necessarily the same. We first assign a privacy level to each trajectory based on the privacy requirement of its moving object. We then partition trajectories into a set of fixed-radius clusters based on the EDR distance. Each cluster is created such that its size is proportional to the maximum privacy level of trajectories within it. We finally anonymize trajectories of each cluster using a novel matching point algorithm. The experimental results show that our approach can achieve a satisfactory trade-off between space distortion and re-identification probability of trajectory data, which is proportional to the privacy requirement of each moving object.