Total Ionospheric Electron Concentration Prediction Using the Inverted Transformer Model

Ionosphere is a space and one of the main sources of interference for Global Navigation Satellite System(GNSS). The ionosphere causes the ionospheric delay phenomenon because the dispersive medium causes the GNSS signal to be refracted. One way to mitigate ionospheric delays is to predict the ionospheric Total Electron Concentration (TEC) in advance. Therefore, the inverted Transformer model is used in this paper to embed solar and geomagnetic activity with spatio-temporal information for TEC prediction. Experiments show that the inverted Transformer model decreases the prediction error by 9.7%-11.4% and 17.5%-26.1 % on a single-point prediction task compared to the Transformer and Informer models. Prediction errors decreased by 46.3% and 27.7% on the Chinese regional point joint prediction task. The used prediction model and dataset effectively predicted the TEC, providing assistance in ionospheric TEC prediction and GNSS receiver interference suppression.