Performance Evaluations of an Equatorial GPS Amplitude Scintillation Detector Using a Machine Learning Algorithm

Yu Jiao, John Hall, Yu (Jade) Morton

Abstract:	This paper applies a machine learning algorithm, called support vector machine (SVM), to autonomously detect equatorial ionospheric scintillation and evaluates its performance using real scintillation data. The SVM detector learns to classify scintillation events based on given training data in the frequency domain. The detector input is the raw signal intensity. Training validation using scintillation data from Ascension Island and Hong Kong shows above 98% accuracy of scintillation event detection. Different training combinations of observation matrices and learning algorithms are investigated to obtain performance measures in terms of receiver operating characteristic curves and confusion matrices. Testing results on data from Ascension Island, Hong Kong, Peru, and Singapore are also presented. The data from the last two locations is not involved in training to demonstrate the general capabilities of the detector.
Published in:	Proceedings of the 29th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2016) September 12 - 16, 2016 Oregon Convention Center Portland, Oregon
Pages:	195 - 199
Cite this article:	Jiao, Yu, Hall, John, Morton, Yu (Jade), "Performance Evaluations of an Equatorial GPS Amplitude Scintillation Detector Using a Machine Learning Algorithm," Proceedings of the 29th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2016), Portland, Oregon, September 2016, pp. 195-199. https://doi.org/10.33012/2016.14554
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In