Automatic modulation classification based on joint feature map and convolutional neural network

Abstract

The authors propose an automatic modulation classification algorithm to discriminate the radar emitter signals exploiting deep learning of convolutional neural network (CNN) based on a joint feature map. The joint feature map including time-frequency map and instantaneous autocorrelation (IA) map is generated to achieve superior classification performance on the identification of both phase modulated signals and frequency modulated signals. Afterwards, a CNN is designed to extract the features of the joint feature map. Training and test samples are designed to evaluate the recognition rate of the CNN. Simulation results show that the joint feature map makes up for the weakness of a single feature map for the classification of frequency modulated signals and phase modulated signals simultaneously, and superior classification accuracy is obtained.