Intelligent Automatic Modulation Classification for Radar Transmitting Signals

Abstract

We propose a technique on automatic modulation classification (AMC) of radar emitter in electronic warfare system. The classification method is based on the perception of instantaneous auto-correlation using a type of 3 dimensional (3D) convolutional neural network (CNN). We can accomplish the AMC without using feature extraction process compared to the conventional methods. The proposed approach is appropriate for the discrimination of several types of signal modulation including single frequency, linear frequency modulation, phase shift keying and frequency shift keying, and its performance is validated via numerical simulation.