Deep Learning-Based Automatic Modulation Classification for Composite Modulated Radar Signal Using Time-Frequency Image

Automatic modulation classification (AMC) of radar signals is a crucial technique in modern electromagnetic warfare. Recently, various AMC methods using deep learning (DL) have been proposed. This paper focuses on an DL-based AMC for composite modulated radar signals, leveraging the time-frequency image (TFI) of the received radar signals. We propose a novel TFI generation method that can reduce distortion to enhance classification performance. The proposed method first generates a grayscale TFI from the received radar signal and reduces its noise via temporal marginalization and the <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mu$</tex>-law function. The resulting image is then standardized and clipped along the frequency axis for pattern enhancement. Through computer simulations on various DL models, we show that the proposed method outperforms the conventional ones in terms of classification accuracy.