A Linear Discriminant Analysis based Modulation Recognition Method for Linear Modulations

In this paper, a new method for signal modulation classification based on Linear Discriminant Analysis for multi-class cases is introduced and its results are compared with two other methods of automatic modulation classification. Higher-order complex cumulants are used as feature vectors in the first place. Then, linear Fisher discriminant analysis maps these vectors to another space to separate different classes efficiently. One versus all support vector machine classifier and kernel Fisher discriminant analysis methods with radial basis function kernel is used. The achieved results in comparison with two research papers show that the proposed method classifies the classes in a shorter time with equal or better accuracy.