Distinguishing Ionospheric Scintillation from Multipath in GNSS Signals Using Bagged Decision Trees Algorithm
Rayan Imam, Fabio Dovis · 2020
This paper presents a machine learning model able to distinguish between ionospheric scintillation and multipath in GNSS-based scintillation monitoring data. The inputs to the model are the average signal intensity, the variance in the signal intensity, and the covariance between the in-phase and the quadrature-phase outputs of the tracking loop of a GNSS receiver. The model labels the data as either scintillated, multipath affected, or clean GNSS signal. The overall accuracy of the model is 96% with 2% miss-detection rate and a negligible false alarm rate for the scintillation class in particular. The gain in the amount of scintillation data is up to 17.5% that would have been discarded if an elevation mask of 30° was implemented.