Effective Feature-Based Automatic Modulation Classification Method Using DNN Algorithm

Abstract

In this paper, we propose an effective feature-based automatic modulation classification (AMC) method using a deep neural network (DNN). In order to classify the modulation type, we consider effective features according to the modulation signals. The proposed method removes the meaningless features that have little influence on the classification and only uses the effective features that have high influence by analyzing the correlation coefficients. From the simulation results, we observe that the proposed method can make the AMC system low complexity.