A Novel Approach for Automatic Modulation Classification via Hidden Markov Models and Gabor Features

In this paper, a novel approach is proposed for classification of pulse amplitude modulated (PAM) and quadrature amplitude modulated (QAM) signals. The automatic modulation classification is the intermediate step between detection and demodulation of the signal. In this paper, we propose a classifier for digital modulated signals such as PAM and QAM that differs with the existing classifiers. The gabor parameters have been used as input features and the proposed classifier uses hidden markov model in conjunction with genetic algorithm (GA). The fitness function for the genetic algorithm is probability of observation sequence given the model. The objective is to maximize the probability of observation using Baum-Welch algorithm and GA. To improve the classification accuracy, classification process has been divided in two phases. Simulation results shows the significant performance improvement while compare with other existing techniques.