LPI Radar Signals Modulation Recognition Based on ACDCA-ResNeXt

For low probability of intercept (LPI) radar waveform identification accuracy (ACC) problem at Signal-to-Noise Ratios (SNRs), an approach based on time-frequency analysis (TFA) and Asymmetric Dilated Convolution Coordinate Attention Residual networks (ACDCA-ResNeXt) is proposed to recognize twelve kinds LPI signals automatically. First, we apply Choi-Williams distribution (CWD), which shows superior performance SNRs, transforming into images (TFI). Then, in order obtain the high-quality TFIs, a series image processing techniques, including 2DWiener filtering, cutting, resize, are used remove background noise redundant frequency bands TFI fixed-size gray scale containing main morphological features TFI. Finally, TFIs input ACDCA-ResNeXt network that can extract learn deep waveforms. Furthermore, fusion loss function, composed soft-label smoothed cross entropy function center improves generalization capability achieves better clustering effect. Experimental results demonstrate that, for waveforms, overall recognition ACC 97.94% when SNR -8 dB.