Train satellite positioning unsupervised deception detection method based on zero deception sample
By employing an unsupervised deception detection method based on zero-deception samples and utilizing a diffusion reconstruction network to learn the time-series graph of satellite signal features, the problems of hardware dependence and data scarcity in existing technologies are solved, achieving efficient and reliable deception detection for train positioning.
Patent Information
- Application Number
- CN202511144931.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-15
- Publication Date
- 2025-12-05
- Estimated Expiration
- 2045-08-15
AI Technical Summary
Existing deception interference detection methods require the introduction of additional hardware in the railway field, increasing system complexity and cost. At the same time, due to the scarcity of real deception interference data, they suffer from insufficient generalization ability and poor reliability, making it difficult to meet the actual needs of train positioning.
An unsupervised deception detection method based on zero-deception samples is adopted. By constructing training and validation datasets that do not contain deception interference, a diffusion reconstruction network is used to learn the time series graph of satellite signal features to detect deception attacks in real time. The reconstruction error is quantified by mean square error and structural similarity index to determine the deception attack.
This technology improves the reliability of train satellite positioning and the accuracy of spoofing detection without requiring additional hardware, thus ensuring the reliability and real-time protection capabilities of train BeiDou positioning.