Multi-Channel Learning with Preprocessing for Automatic Modulation Order Separation

Abstract

Automatic modulation classification (AMC) with deep learning (DL) based methods has been studied in recent years and improvements have been shown in many studies; however, it has been difficult to design a classifier that can distinguish modulation orders such as 16-QAM and 64-QAM, with high accuracy. In this study, the distinction performance of 16-QAM and 64-QAM modulation orders increased by feeding the features obtained during the preprocessing stage to the multi-channel convolutional long short-term deep neural network (MCLDNN). Simulation results indicate performance improvements, particularly at the low SNR region. Furthermore, the proposed method can be extended for the separation of other orders of QAM and other digital modulations.