Real time GPS Positioning of LEO Satellites Mitigating Pseudorange Multipath through Neural Networks

Abstract

A method for real-time positioning of LEO satellites using dual frequency GPS receivers is presented. It is based on an a priori ground estimation of a pseudorange multipath map computed by means of a Self-Organizing Map neural network algorithm. The generated map characterizes the multipath environment of the satellite. This a priori estimation allows a real time correction of the pseudorange observables onboard the LEO satellite with a number of parameters affordable for space applications in terms of CPU and memory usage. The novelty of the approach consists of the use of neural networks to reduce the number of parameters and the use of a hybrid offline-online method. Precise IGS clocks and orbits have been used to measure the impact of these corrections in the navigation solution. Improvements in 3D positioning error of about 40%-50% for SAC-C (obtaining errors $90cm) and 25%-35% for CHAMP (obtaining errors $70cm) are demonstrated.

A Systematic Review of Machine Learning Techniques for GNSS Use Cases

“…This is beneficial to many end-users applications as it provides faster access to the required orbit solutions [145]. Machine learning for orbit determination of LEO satellites has been implemented by some studies [146][147][148][149]. In [146,149], Time delay neural network (TDNN) which is a type of Feed Forward Neural Network (FFNN), and LSTM are used for simultaneous tracking and navigation with LEO satellites.…”

“…In [146,149], Time delay neural network (TDNN) which is a type of Feed Forward Neural Network (FFNN), and LSTM are used for simultaneous tracking and navigation with LEO satellites. A neural network is implemented in [147,148] to mitigate GNSS multipath for LEO positioning applications. It was noticed that very limited studies has been done on the use of ML for POD of LEO satellites.…”