Use of Supervised Machine Learning for GNSS Signal Spoofing Detection with Validation on Real-World Meaconing and...

Area of Science:

• Navigation Systems
• Signal Processing
• Machine Learning

Background:

• Global Navigation Satellite System (GNSS) open service signals are susceptible to spoofing and meaconing attacks.
• These attacks can lead to hazardous misleading information in safety-of-life applications without detection.
• Current detection methods have limitations in identifying manipulated GNSS data.

Purpose of the Study:

• To present a novel method for detecting spoofed or meaconed GNSS signals.
• To enhance the integrity and reliability of GNSS data for safety-critical users.
• To leverage supervised machine learning for robust GNSS signal authentication.

Main Methods:

• Cross-correlation monitoring of multiple GNSS observables and measurements.
• Utilizing statistically significant GNSS data as input for machine learning algorithms.
• Training and validation using laboratory-generated and real-world spoofing/meaconing datasets.

Main Results:

• Demonstrated the effectiveness of the proposed method in detecting spoofed signals in controlled experiments.
• Validated the supervised machine learning algorithms on diverse real-world spoofing and meaconing datasets.
• Achieved reliable detection of manipulated GNSS signals, mitigating hazardous misleading information.

Conclusions:

• The cross-correlation monitoring and machine learning approach provides a viable solution for GNSS spoofing and meaconing detection.
• This method enhances the security and integrity of GNSS open service signals.
• The findings contribute to improving the safety and reliability of safety-of-life applications reliant on GNSS.