Characterization of Ionospheric Scintillation Using Deep Learning Models

Rubem Vasconcelos Pacelli, Rodrigo de Lima Florindo, Felix Antreich, André Lima Ferrer de Almeida, Angela Aragon-Angel, Adrià Rovira Garcia

Peer Reviewed

Abstract:	Ionospheric scintillation is a phenomenon that affects the quality of Global Navigation Satellite System signals, leading to degradation in positioning accuracy and reliability. This paper presents the analysis of discriminative deep neural networks (DNNs) for the classification of ionospheric scintillation time series. The DNNs are trained on a dataset generated using a simulation framework based on a phase screen model, which allows for the generation of scintillation time series data. The framework incorporates several key input spaces, including irregularity parameters, receiver positions, and zonal drift velocities, to simulate realistic scintillation events. The DNN architecture is based on a convolutional neural network and a multilayer perceptron, which allows for the extraction of temporal features from the input data and classification of the scattering regime as weak or strong. Additionally, new metrics from the time series are introduced to enhance the characterization. The model is trained and evaluated on the generated dataset, achieving high classification accuracy, precision, recall, and F1 score. The results demonstrate the effectiveness of the proposed framework and model for the classification of scintillation time series.
Published in:	Proceedings of the 38th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2025) September 8 - 12, 2025 Hilton Baltimore Inner Harbor Baltimore, Maryland
Pages:	3339 - 3352
Cite this article:	Pacelli, Rubem Vasconcelos, Florindo, Rodrigo de Lima, Antreich, Felix, de Almeida, André Lima Ferrer, Aragon-Angel, Angela, Garcia, Adrià Rovira, "Characterization of Ionospheric Scintillation Using Deep Learning Models," Proceedings of the 38th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2025), Baltimore, Maryland, September 2025, pp. 3339-3352. https://doi.org/10.33012/2025.20390
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In