Implementation of IoT analytics ionospheric forecasting system based on machine learning and ThingSpeak

Abstract

Nowadays, using the Internet of Things (IoT), several real-time forecasting systems have been developed. The primary challenge of this system is to utilise an appropriate prediction model that can predict various space weather parameters as accurately as possible. In this study, an ionospheric IoT analytical system with variational mode decomposition (VMD) based on kernel extreme learning machine (KELM) is proposed. The ionospheric signal delay/total electron content (TEC) data from Continuous Reference Stations (CORS) Port Blair (2.03°N, 165.25°E, geomagnetic), Bengaluru (4.40°N, 150.77°E, geomagnetic), Koneru Lakshmaiah Education Foundation (KLEF) - Guntur (7.50°N, 153.76°E; geomagnetic) and Lucknow (17.98°N, 155.22°E; geomagnetic) are used for the analysis during the period of 2015. The ionospheric signal delays of four CORS are computed from ThingSpeak (IoT) with the channel ID and the Application Programming Interface key. ThingSpeak data is given to the ionospheric forecasting model (VMD-KELM). The results predicted from the proposed model are able to achieve the faster training process and obtain a similar accuracy to that of the VMD-artificial neural network. The proposed VMD-KELM application is adopted when a cloud-based forecasting system requires fast learning speed and good accuracy. As a result, the cloud paradigm offers the possibility without web development skills or highly specific statistics.