Density clustering-based optimization model for trajectory data publication

Trajectory data is vital for enhancing location-based APP services. However, releasing such data without proper privacy safeguards exposes user privacy risks. Currently, most differential privacy methods focus on static data, while existing obfuscation techniques for locations suffer from limited usability and inadequate privacy protection. To tackle these issues, we introduce DCT-DP, an optimized model for trajectory data publication based on density clustering. DCT-DP first segments trajectory temporal attributes using k-means clustering. Then, it utilizes a novel density clustering algorithm, DENCLUS, to categorize spatial attributes. Following this, it filters out abnormal trajectories and applies Laplace noise to protect privacy before publishing. Our experimental evaluation demonstrates that DCT-DP outperforms existing methods, offering superior data utility while ensuring stronger privacy protection.