Investigation of Machine Learning Approaches for Retrieval of Soil Moisture Using NavIC Signal

Navigation with Indian Constellation (NavIC) is India's own navigation system to cater the navigational and surveying need of the Indian subcontinent. The navigation signals transmitted by the NavIC satellites are available round the clock and scientists have found its use beyond the navigation and surveying purposes. The study of soil moisture with navigation signal is one of the most recent applications which has attracted the research community. The ground-based receiver collects the multipath signal, which includes the direct signal and reflected signal from the soil surface, to calculate the multipath phase, which is sensitive to soil moisture. To determine the soil moisture with multipath phase, this work investigates the application of machine learning (ML) technique. The investigation has utilized the support vector machine (SVM), decision tree, random forest (RF), and linear regression techniques. When predicting soil moisture, random forest performs better than other algorithms.