Segment Clustering Based Privacy Preserving Algorithm for Trajectory Data Publishing

Abstract

In recent years, privacy preserving data publishing has become a hot research point in the area of privacy preserving. The (k, δ)-anonymity model exploited the inherent uncertainty of the trajectory acquisition system, but could not consider the case that the uncertain threshold of trajectory was variable in practical applications. Therefore, the (k, δ)-anonymity model was improved, and a (k, Δ)-anonymity model with variable threshold was proposed. Furthermore, in the traditional cluster-constraint based trajectory anonymous algorithm, the whole trajectory was taken as the basic unit for clustering. Although the probability that attacker can identify the trajectory of a specific user can be reduced to 1/k, the anonymous group was vulnerable to the re-clustering attack as the anonymous group lacks diversity. Aiming at this problem, a segment clustering based trajectory privacy preserving algorithm was proposed. The trajectories were partitioned into segments based on the principle of Minimum Description Length. And then, these segments were made anonymous based on the cluster-constraint strategy. Simulation results show that the proposed method can improve the security, and have better performance in terms of data quality and data availability.