A privacy-preserving trajectory data synthesis framework based on differential privacy

Abstract

Trajectory data contains spatio-temporal information about a large number of moving objects, and analysis of this data can yield a lot of valuable information to support a variety of application requirements related to moving objects. However, publishing unprotected trajectory data carries the risk of revealing personal privacy. In the existing approaches, the utility of published trajectory data is sacrificed for privacy protection. In this paper, we propose a differential privacy mechanism DP-Std for data publication in synthetic trajectory databases. first, the DP-Std mechanism constructs a density- and distance-aware adaptive grid structure for discrete geographic regions, and adds noise in the process to satisfy differential privacy. Second, DP-Std extracts feature distributions to ensure the utility of synthetic trajectories. Third, DP-Std constructs a mobility model that samples intermediate location points. And then the trajectory dataset is synthesized by differential privacy distribution and mobility model. Finally, we conduct experiments on two datasets to evaluate our framework. The results show that DP-Std outperforms other feature-based trajectory synthesis methods in terms of data utility and achieves a trade-off between privacy and utility with strict privacy protection.