Vehicle trajectory data de-identification method and system based on differential privacy
By dynamically and differentially allocating privacy budgets through geographic grid division and multidimensional feature analysis, the problem of privacy budget mismatch in existing technologies is solved, achieving efficient desensitization of vehicle trajectory data while balancing the protection of high-risk nodes and the data availability of low-risk nodes.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SEEWORLD TECH CO LTD
- Filing Date
- 2026-05-13
- Publication Date
- 2026-07-24
AI Technical Summary
Existing technologies cannot balance the key defense of high-risk nodes with the data availability of low-risk nodes in the desensitization of vehicle trajectory data, resulting in a mismatch of privacy budgets, failing to effectively protect highly sensitive stop points, and compromising the path continuity and macro-statistical availability of the trajectory.
By extracting the dwell coordinate set based on the geographic grid division, calculating multidimensional dwell features, and dynamically and differentially allocating privacy budgets according to sensitive categories, differential privacy noise is injected into the passing and dwell grid cells respectively to achieve differentiated noise addition processing.
By precisely matching protection strength with actual sensitivity requirements, the protection of highly sensitive areas is enhanced, while the data availability of low-sensitivity areas is preserved, thereby improving the overall security and macroscopic analysis availability of the desensitized trajectory.
Smart Images

Figure CN122197078B_ABST