A Convolutional and Transformer Based Deep Neural Network for Automatic Modulation Classification

Shanchuan Ying,Sai Huang,Shuo Chang,Zheng Yang,Zhiyong Feng,Ningyan Guo

DOI: https://doi.org/10.23919/jcc.ja.2022-0580

2023-01-01

China Communications

Abstract:Automatic modulation classification (AMC) aims at identifying the modulation of the received signals, which is a significant approach to identifying the target in military and civil applications. In this paper, a novel data-driven framework named con-volutional and transformer-based deep neural network (CTDNN) is proposed to improve the classification performance. CTDNN can be divided into four modules, i.e., convolutional neural network (CNN) backbone, transition module, transformer module, and final classifier. In the CNN backbone, a wide and deep convolution structure is designed, which consists of 1 × 15 convolution kernels and intensive cross-layer connections instead of traditional 1×3 kernels and sequential connections. In the transition module, a 1 × 1 convolution layer is utilized to compress the channels of the previous multi-scale CNN features. In the transformer module, three self-attention layers are designed for extracting global features and generating the classification vector. In the classifier, the final decision is made based on the maximum a posterior probability. Extensive simulations are conducted, and the result shows that our proposed CTDNN can achieve superior classification performance than traditional deep models.