A Privacy Protection Scheme to Trajectory Synthesis Using Differential Privacy

Trajectory data records the movement path of moving objects in space, which plays a vital role in geographical location analysis and planning, personalized service and recommendation.However, if the original trajectory data is not well protected or released directly, personal privacy is easy to be disclosed.Therefore, to protect privacy security and ensure the usability of trajectory data, effective protection measures are essential.In this paper, we propose Differential Privacy Trajectory Synthesis (DPTS), a privacy protection scheme that uses differential privacy to synthesize trajectory data.We use differential privacy technology and Markov models to generate synthetic trajectory datasets.The trajectory is first discretized in geographic space with the aim of creating discrete states for the Markov model.Secondly, the Markov mobility model is constructed, and the trajectory features are extracted by adding Laplace noise to disturb the intra-trajectory mobility, trajectory length, and trajectory trip distribution.Among them, the log-odds conversion method is proposed to calculate the Laplace noise probability matrix.Then, the trajectory synthesis algorithm is proposed to synthesize the trajectory.We implemented experiments on three datasets respectively to obtain synthetic trajectory datasets, and evaluated the effectiveness of our method on the synthetic datasets and the original datasets.The experimental results indicate that the trajectory data synthesized by DPTS has higher privacy and utility than the compared state-of-the-art methods.Compared with the comparison method, DPTS improves data utility by 25.6% overall.