Date:
Type:
Publication:
Author(s):
Magnetic Aircraft Navigation (MagNav) systems leveraging the Earth’s crustal magnetic field demonstrate resilience against traditional navigation challenges such as weather disruptions, jamming, and spoofing. These systems provide a critical fail-safe for GPS-based and inertial-based navigation. However, significant magnetic interference from the aircraft, onboard electronics, and other natural magnetic sources makes it challenging to detect the target crustal field signal for navigation. While most noise sources can be modeled and characterized, rapidly varying magnetospheric-ionospheric fields present a challenge for MagNav observations.
In this study, we explore characterizing and correcting for magnetospheric-ionospheric fields using various Artificial Intelligence (AI) models that incorporate external field inputs during flights from open-source geomagnetic models. Our study explores the use of deep learning algorithms to approximate nonlinear temporal signals by incorporating magnetospheric-ionospheric field signals. These models are compared against traditional machine learning algorithms in terms of their ability to isolate the target crustal field signal used for navigation.
Performance trends observed in the best-performing AI models provide insights into the spatio-temporal accuracy of external field model outputs under various solar wind and solar activity conditions. These trends also highlight the necessary magnetospheric-ionospheric field characterization from state-of-the-art models required to enhance MagNav robustness.
