Exploring the correlation between ionospheric scintillation and GNSS positioning error near the magnetic equator using machine learning techniques

Abstract

Ionospheric scintillations are a common phenomenon in the equatorial ionosphere. This phenomenon directly affects the position estimated by GNSS receivers degrading the quality of the radio signals; however, the quantification of the positioning error contributed by the ionosphere over the Peruvian sector has not been studied in detail. In this work, algorithms are being implemented that will allow us to identify and classify amplitude scintillation (S4) levels, we have worked with data from the Huancayo Observatory for the period December 2016-February 2017 obtained from LISN, this data has been plotted to analyze the spatial and temporal occurrence, and to analyze the occurrence of S4 as a function of other space weather variables obtained from OMNI2. The machine learning algorithms were decision tree, Support Vector Machine (SVM), Neuronal Network (NN). Decision tree was implemented as a filtering method, support vector machine for clustering and neuronal network to generate time series in forecasting. This paper shows the initial part of an investigation that aims to correlate qualitatively and quantitatively the occurrence of amplitude scintillations (S4) with errors in the position estimation of GNSS receivers, once the correlation between S4 and position error has been quantified, it may be possible to predict the error by predicting S4.