A Data Publishing Method for Trajectory Privacy Classification Based on Differential Privacy

The traditional privacy protection method of trajectory data release protects the whole trajectory to the same degree, resulting in an unjustifiable allocation of the privacy budget and a diminishment in data accessibility. A technique is put forth to partition the privacy level of the trajectory based on the visitation frequency and duration, catering to the privacy preservation requirements of users with diverse data sensitivities. Density-based Spatial Clustering of Applications with Noise (DBSCAN) clustering is employed to categorize the trajectory points within high-density clusters into identical clusters, thus forming distinct trajectory clusters. The trajectory is segmented using the standard deviation to ensure uniform distribution of trajectory segments. This process streamlines the dataset, extracts distinctive behavioral patterns, and mitigates the spatiotemporal intricacies associated with trajectory data processing. A noisy trajectory segment prefix tree is assembled, and the privacy budget is allocated based on the weightage of the trajectory’s privacy level and the tree’s height. A Markov chain is introduced to constrain the magnitude of noise injected into the data. Experimental results substantiate the efficacy of the algorithm proposed in this paper in balancing data availability and privacy preservation.