Personalized Location Intelligence Privacy Protection Technology Based on Differential Privacy Mechanism in Different Data Publishing Scenarios

In the current era of mobile Internet, personalized location services have significantly enhanced user experience. However, they have concomitantly given rise to substantial challenges related to the protection of location privacy. The study aims to develop a location privacy protection model for static and dynamic data publishing scenarios by deeply analyzing the differential privacy mechanism and its application on mobile aggregated data features. The study designs a protection model using diverse perturbation mechanisms for static data and constructs a set of protection models for dynamic data by combining prediction, adaptive adjustment, and data grouping and merging. The experiments were validated using the Chengdu cab trajectory dataset (real data) and the simulated synthetic population movement dataset (simulated generated data). The experimental results showed that the static model successfully reduced the trajectory recovery rate to 18% and achieved the lowest mean absolute error of 0.8 and the lowest mean relative error of 0.45 on the simulated dataset. The dynamic model achieved a minimum mean absolute error of 0.047 and a minimum mean relative error of 20 with a fixed privacy budget. The above results demonstrate the potential of the two models in improving the efficacy of location privacy protection and point out the direction for the future development of privacy protection techniques in personalized location services.