History trajectory privacy-preserving through graph partition

Abstract

With rapid development of positioning techniques and location based services (LBS), locations and traces of moving objects are collected by service providers, the data will then be published for novel applications. Although analyzing and mining trajectories is useful for mobility-related applications, new challenges of trajectory privacy leakage arise accordingly. Trajectories contain rich spatio-temporal history information that may expose users' whereabouts and other personal privacy. At present, trajectory k-anonymity which aims at anonymizing k trajectories together on all sample points is one of the most popular techniques to protect trajectory privacy. The challenge lies in how to find trajectory k-anonymity sets. In this paper, a trajectory graph is constructed to simulate spatial relations of trajectories, based on which we propose to find trajectory k-anonymity sets through graph partition, which is proven NP-complete. We then propose a greedy partition method to find trajectory k-anonymity sets, as well as yielding low information loss. We run a series of experiments on both real-world and synthetic datasets, the results show the effectiveness of our method.