Decision tree based GNSS positioning

Position information and time information provided by GNSS (Global Navigation Satellite System) are positively used. The accuracy of GNSS information is a very important factor for future ICT based systems such as an autonomous driving car, 5G wireless system etc. To meet such a demand, this work applies a machine learning technique to GNSS positioning and shows the feasibility of machine learning based GNSS positioning. As one of the advantages, our proposed system can make full use of current GNSS receiver system, that is, it does not need the modification of current device except for the signal processing architecture. Simulation results show that the proposed decision tree based GNSS positioning can enhance both accuracy and continutity of positioning compared to the conventional technique and the random forest based GNSS positioning can further improve both accuracy and continutity of positioning. © 2021 The Institute of Electrical Engineers of Japan.