Comparative Performance Study of Linear and Gaussian Kernel SVM Implementations for Phase Scintillation Detection

Abstract

The aim of this paper is to analyze the design of support vector machine (SVM)algorithm that belongs to the class of supervised machine learning algorithms for phase scintillation detection and to discuss the performance comparison of linear and Gaussian kernel implementations by considering the design parameter's effects. The algorithm processes the phase scintillation indices computed for GPS L1 signals through the designed linear and Gaussian kernel SVM models. The study is based on the real GNSS signals which are affected by phase scintillations, collected at South African Antarctic research base (SANAE IV).