A modulation classification based on SVM

Abstract

In this paper, we focused on the problem of automatic modulation classification of digital signals. Several useful characteristic parameters which can be used for modulation analysis are extracted from spectral correlation, for different types of modulated signals have different power spectral density functions. A density estimation approach based on Support Vector Machine (SVM) is developed. Also, a kind of Bayesian classifier is constructed using the estimated probability density. The experiment results demonstrated that the Bayesian classifier designed in this paper outperformed the traditional SVM classifier in the aspect of classification accuracy and training speed.