Fault-Robust GPS Spoofing Mitigation with Expectation-Maximization

Ashwin Vivek Kanhere, Grace Gao

Abstract: The possibility of measurement faults and spoofing attacks poses real-world risks to accurate and safe localization using GPS measurements. Measurement faults introduce additive biases in individual measurements that might be inconsistent across measurements while spoofing attacks introduce consistent additive biases in all measurements. Both of these effects introduce errors in the localization solution estimated using these affected measurements. Recognizing these risks, researchers have developed solutions to mitigate these spoofing attacks and measurement faults, usually individually, by using redundant measurements or explicitly modelling and estimating the attack and fault magnitudes. When these localization methods model spoofing or faults individually, they can raise a large number of false spoofing alerts in the presence of faults or incorrectly incorporate spoofing attacks when accounting for faults. However, as both spoofing attacks and faulty measurements can be encountered in the real world, there is a need to mitigate both jointly when performing GPS localization. In this work, we propose an expectation-maximization (EM)-based method for jointly mitigating the effects of spoofing attacks and measurement faults during localization. During the expectation step, in a two-step process, we first estimate the likelihood that the GPS measurements are spoofed at a particular time instant using sensor level redundancy. We then estimate the likelihood that individual GPS measurements are faulty at that time instant using measurement-level redundancy. Finally, during the maximization step, both these likelihood estimates are combined to de-weigh GPS measurements that are used for localization. We also highlight a particular implementation of our proposed method. Our method estimates the likelihood of GPS spoofing using M-estimation and the likelihood of measurement faults along with state estimate with a factor graph optimization framework with switchable constraints (SC-FGO). We experimentally validate our proposed approach in simulations, comparing to three different baseline implementations, to show that our method successfully mitigates the effects of both spoofing and faulty measurements.
Published in: Proceedings of the 36th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2023)
September 11 - 15, 2023
Hyatt Regency Denver
Denver, Colorado
Pages: 3815 - 3828
Cite this article: Kanhere, Ashwin Vivek, Gao, Grace, "Fault-Robust GPS Spoofing Mitigation with Expectation-Maximization," Proceedings of the 36th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2023), Denver, Colorado, September 2023, pp. 3815-3828. https://doi.org/10.33012/2023.19366
Full Paper: ION Members/Non-Members: 1 Download Credit
Sign In