Successive Trajectory Privacy Protection with Semantics Prediction Differential Privacy

Area of Science:

• Computer Science
• Data Privacy
• Location-Based Services

Background:

• Publishing trajectory data is crucial for location-based services but poses privacy risks.
• Existing differential privacy methods for trajectories often neglect user location preferences and semantic context.
• A key challenge is balancing privacy protection with service quality in differential privacy for trajectory data.

Purpose of the Study:

• To propose a novel differential privacy scheme for trajectory data.
• To enhance trajectory data privacy while maintaining data availability and service quality.
• To address the limitations of existing schemes by incorporating semantic location and predictive analysis.

Main Methods:

• Trajectory data is transformed into a prefix tree structure to ensure differential privacy.
• Semantic sensitivity and location check-in frequency are used to calculate position sensitivity.
• Privacy levels are classified, and privacy budgets are allocated based on these levels.
• A Markov chain predicts attack probabilities to further adjust budget allocation and improve utilization.

Main Results:

• The proposed scheme effectively protects trajectory data privacy.
• Data availability is maintained, ensuring utility for location-based services.
• The scheme successfully balances the privacy budget with service quality.

Conclusions:

• The semantics- and prediction-based differential privacy scheme offers an improved approach to publishing trajectory data.
• This method enhances privacy protection without significantly compromising the usability of the data.
• It provides a robust solution for the trade-off between privacy and service quality in trajectory data publication.