Analysis of Ionospheric Scintillation Detection Based on Machine Learning

摘要

Ionospheric scintillation, characterized by deep fluctuation of signal intensity on amplitude and phase, poses threats to the global navigation satellite system (GNSS) tracking performance, even causing degradation in position and navigation in accuracy, integrity and continuity. To achieve higher performance of ionospheric scintillation detection, a method of machine learning is implied to attain automatic detection. Three levels of scintillation intensities and two types of scintillation features are investigated and compared. Based on data from São José dos Campos, Brazil (23.2S, 45.9W), training validation and prediction results show 99% accuracy approximately and above 96%, respectively. The analysis will provide useful guidance on ionospheric scintillation detection to make of great signification on design of advanced GNSS receiver and research of atmospheric layer and space weather.