Hybrid Deep Learning Model (LSTM-CNN) Application for GPS-Based Ionospheric TEC Forecasting

D. Venkata Ratnam Senior Member

Authors

D. Venkata Ratnam Senior Member Author

Abstract

Deep learning algorithms have shown great promise in forecasting low latitude ionospheric disturbances, such as delays in Global Positioning System (GPS) signals.This letter explores the application of deep learning models, specifically Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and a hybrid model combining LSTM with Convolution Neural Network (CNN), to forecast ionospheric delays for GPS signals. The data used for training and testing the deep learning models is the vertical Total Electron Content (VTEC) time-series data estimated from GPS measurements collected at Bengaluru, Guntur, and Lucknow GPS stations.Among the various deep learning forecasting algorithms for the ionosphere, the LSTM-CNN model stands out with the best performance. It achieves a minimum root-mean-square error (RMSE) of 1.5 Total Electron Content Units (TECUs) and a high degree of accuracy with an R2 value of 0.99. These results demonstrate the effectiveness of the LSTM-CNN model in forecasting ionospheric delays for GPS signals.