A Machine Learning Approach for GPS Code Phase Estimation in Multipath Environments
Mohamad Orabi, Joe Khalife, Ali A. Abdallah, Zaher M. Kassas, Samer S. Saab
Abstract
A neural network (NN)-based delay-locked loop (DLL) for multipath mitigation in Global Positioning System (GPS) receivers is developed. The NN operates on equally-spaced samples of the autocorrelation function. The NN is trained using a statistical distribution model that takes into consideration multipath time delay and power attenuation. The performance of the proposed method is compared numerically and experimentally with three other conventional techniques: conventional early-minus-late DLL, narrow correlator, and high resolution correlator. It is demonstrated that the NN-based DLL produces smaller code phase root mean squared error compared to the three conventional techniques in high multipath environments.