Applicability of GNSS Pseudorange Residual Error Mitigation Model to Different Types of Receivers

Global Navigation Satellite Systems (GNSS) are crucial in the modern world and provide precise positioning, navigation, and timing services worldwide. Over the past decade, GNSS found application in a wide range of fields, and critical infrastructure elements are highly dependent on GNSS technology, therefore, GNSS is evolving and modernizing. One of the main GNSS characteristics is positioning accuracy. There are many different approaches to improving positioning accuracy. Previously a method for compensating residual pseudorange error was developed. This method is based on machine-learning ability to find the relationship in data and allows one to improve positioning accuracy. A large number of experiments with eight navigation receivers and two navigation antennas were conducted and the results are presented in the paper. Navigation measurements were collected over a 14-day interval for eight receivers, pseudorange reference values and residual pseudorange errors were calculated, individual machine learning models were trained for the presented receivers and two metrics (root mean square error and explained variance) for these models were evaluated. The obtained results of model training are worse than previous studies, but the ability of the method to work with different navigation equipment was successfully confirmed during the experiment, including navigation receivers manufactured in Russia. During the experiment, Septentrio PolarX5 showed the best result, although it runs in “user-friendly‘” conditions, while the worst result showed Javad Sigma <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$G$</tex> 3T, although its ability is almost the same as for the other receivers.