Differentially Private Data Publishing of Trajectory Synthesis Based on Generalization and Probability

With the advancement of information technology, the value of data has further emerged. Trajectory data, being a type of massive data, has emerged as a valuable asset in enterprises and a driving force for innovation. However, privacy issues are also increasingly prominent. As a result, developing effective methods for protecting the privacy of trajectory data has become a research hotspot. However, most existing methods ignore temporal attributes and spatial distribution characteristics of trajectory data, resulting in loss of important information and reduced efficiency. To improve on the method, a new differentially private trajectory-data publishing algorithm, differentially private trajectory-data publishing based on generalization and probability (TPGP), is proposed in this work. The algorithm has three stages and generates a synthetic trajectory dataset. The first stage pre-processes trajectories, by performing time splitting and Hilbert space partitioning on the compressed trajectory data, without ignoring the time attribute. In the second stage, Laplace noise is added to obtain two key pieces of statistical information: the noisy counts of generalized trajectory and the noisy Markov transition probability with time attribute. The third stage generates and releases synthetic trajectories using the two pieces of statistical information obtained in the second stage. Experimental results indicate that the proposed TPGP scheme has significant advantages over existing methods in terms of ensuring data privacy while improving data utility.