Recognition and separation of radar aliasing signals based on CNN and SVDD

Abstract

The identification and separation of aliasing signals play a key role in radar electronic countermeasures. Due to the complex and diverse waveforms of radar radiator signals and the characteristics of rapid parameter changes, as well as the uncertainty of the aliasing types of unknown signals and the uncertainty of signal composition, the identification and separation of radar aliasing signals are facing challenges. In response to the above challenges, a recognition architecture based on Convolutional Neural Network (CNN) and Support Vector Data Description (SVDD) was designed to identify and separate aliasing signals of unknown signal types. First, the CNN network extracts the features of the single-signal image and submits it to the SVDD classifier for training, so that the aliasing signal can be filtered from the unknown signal type (including the aliasing signal and the single signal). Secondly, we put the filtered aliasing signals into the CNN network for identification and classification to obtain the signal composition and aliasing type. Finally, according to the identified signal aliasing type and signal type, combined with MATLAB, different methods are selected to separate the aliasing signal. Simulation experiments prove that the recognition accuracy of this method is higher than that of traditional methods.