Improving RTDGPS accuracy using hybrid PSOSVM prediction model

Abstract

Users of Global Positioning System (GPS) require more accuracy from positioning service of this system. Accurate and real time positioning is realized through implementing Real Time Differential GPS (RTDGPS). Accuracy of this system depends on successive and punctual transmission of Pseudo Range Corrections (PRCs) from Reference Station (RS) to user station. Corrections are transmitted using Radio Technical Commission for Maritime services (RTCM) protocol. To compensate signal absence and transmission delay of differential corrections which is created while transferring, processing and reception, a hybrid prediction algorithm combined of Support Vector Machine (SVM) and Particle Swarm Optimization (PSO) has been implemented in RS. Corrections are predicted using SVM while, PSO is used for determination and optimization of the basic parameters in SVM algorithm. A Low Cost GPS receiver is used in RS. Prediction time step was five seconds. Simulation results show that the total Root Mean Square (RMS) prediction error for PSOSVM is less than 0.126 m for one step ahead and 0.46 m for 10 step ahead predictions. Experimental results indicate that by employing PSOSVM, accuracy of RTDGPS has been improved.