Radar PRI Modulation Type Recognition Based on GAF-SE-CNN

The Pulse Repetition Interval (PRI) parameter of a radar is closely related to the radar waveform, working mode, and system resource scheduling, et. al. Correctly identifying the PRI modulation type is of great significance for inferring the technical system, working platform, and behavioral state of the radar emitters. This paper proposes a PRI modulation recognition algorithm based on GAF-SE-CNN to address the problem of low recognition accuracy of existing algorithms under non-ideal conditions with high proportions of lost pulses and spurious pulses. This method first uses the gramian angle field (GAF) algorithm to process PRI sequences of different categories into two-dimensional images with greater feature differences. Secondly, the squeeze and excitation convolutional neural network (SE-CNN) model is constructed to make important features play a leading role in PRI modulation identification. Finally, the processed 2D images are input into the SE-CNN model for training and testing. The experimental results show that the proposed method has good robustness; even with lost pulse and spurious pulse ratios as high as 60%, the recognition accuracy is still very high.