Comparison between Wavelet and Radial Basis Function Neural Networks for GPS Prediction

Abstract

Neural networks are complex nonlinear models;this characteristic enables them to be used in nonlinear system modeling and prediction applications.The estimation and prediction are importantroles in the communication system.The proposed approach based onthe Wavelet Neural Networks (WNNs)usesmorlet as an activation function in thehidden layer of the wavelet neural network,while the Radial Basis Function Neural Networks (RBFNNs)usebasis functionthat can be calculated as a Gaussian function. In this paper,a comparisonbetween the performance ofWavelet Neural Networksand Radial Basis Functionfor GPS prediction is presented.The comparison results(usingMATLAB programming)presentthat the Wavelet Neural Networks method has a great approximation ability, suitability and more stable in Global Positioning System (GPS)prediction than the Radial Basis Function Neural Networks,were highly effective predictions for accurate positioning and RMS errors are 0.05meter after using of Wavelet Neural Networks prediction.