Prediction of VTEC Based on NavIC/GPS Data by Using Machine Learning Algorithms and Comparison with IRI Models

Abstract Total Electron Content (TEC) is utilized for the estimation of Ionospheric delay. Accurate prediction of TEC helps correct errors in position estimation using range measurements. TEC is influenced by several factors, such as location, time, solar radiation, geomagnetic index, and season. In this research, machine learning algorithms such as DeepAR, DLinear, and FFNN are used to forecast the Vertical TEC. The developed algorithms are evaluated using L5- and S-band pseudorange measurements from an operational NavIC receiver at ACS College of Engineering (ACSCE), Bangalore, India. In order to assess the algorithms, NavIC TEC for the ACSCE station is predicted and analyzed. The NRMSE of DeepAR is 0.0.225, MASE is 1.468 and SMAPE is 18%. The FFNN yielded an NRMSE of 0.136, SMAPE of 9.51% and MASE of 0.839. The DLinear provides an NRMSE of 0.157, MASE of 0.961 and SMAPE of 11.57%. Apart from this, the VTEC predicted by the DeepAR, FFNN, and DLinear algorithms are compared with the IRI-2016 and IRI-2020 models, and it is found that FFNN gives better results compared to DeepAR, DLinear, IRI-2020, and IRI-2020.