Modeling of GPS SPS Timing Error using Multilayered Neural Network

GPS is not only an accurate navigation system; it also delivers time with unprecedented accuracy. In this paper, a multilayered neural network (MNN) based approach for forecast and improvement of GPS standard positioning service (SPS) timing error is presented. The proposed MNN is trained using back-propagation (BP) and extended Kalman filter (EKF) training algorithms. The performance of these proposed MNNs is demonstrated by showing its effectiveness in GPS timing error prediction of a low cost GPS receiver. The tests results on the collected real data show that GPS timing error RMS can reduce from 300 nsec and 200 nsec to less than 120 nsec and 43 nsec by using MNN prediction, before and after SA, respectively. The experimental results emphasize that performance of MNN based on the EKF training algorithm is better than BP