Privacy Preserving Serial Publication of Trajectory Data

Abstract

Sharing trajectory data with researchers or any organization is very challenging due to individual privacy risk. Specially it becomes more challenging when trajectory data have to share serially after a specific interval. All the existing methods are not applicable for serial publication. Methods for serial publication is required because of dynamic behaviour of trajectory data. In a serial publication of trajectory data, two types of privacy guards must be provided. Firstly, privacy guards is to be used in each release. Secondly, there must be a privacy guard between two releases, so that individual trajectory is not identified. To fulfill these two objectives, we propose a model that consists of space shifting and spatiotemporal points clustering. In our model, two types of clustering approaches have been used. Each cluster meets the concept of k-anonymity that prevent record linkage attack in each release. After applying different clustering algorithm in each release, experimental result shows that our propose approach is able to mitigate intersection attack.