Function Recognition Of Multi-function Radar Via CNN-GRU Neural Network

• Main Authors: Chen, Hongyu, Feng, Kangan, Kong, Yukai, Zhang, Lidong, Yu, Xianxiang, Yi, Wei
• Format: Conference Proceeding
• Language: English
• Subjects: CNN-GRU; Feature extraction; function recognition; intercepted pulse stream sequence; Logic gates; MFR; Neural networks; Radar equipment; Reconnaissance; Simulation; Time-frequency analysis
• ISSN: 2155-5753

In the field of cognitive electronic reconnaissance, recognizing the function (A variety of work modes arranged in temporal sequence) of the multi-function radar (MFR) is critical for electronic warfare equipment to develop effective countermeasures. However, research in this field is still very lack. Therefore, this paper proposes a convolutional neural network and gated recurrent units (CNN-GRU) to achieve MFR function recognition. The one-dimension convolutional neural network (1D-CNN) structure can be adapted to significantly reduce the computation time when processing a long input sequence, as well two 1D-CNNs are utilized to extract the higher-order sequential features of pulse repetition frequency (PRF) and pulse width (PW) in intercepted pulse stream sequence, respectively, while the GRU learns the higher-order sequential features to output the recognition results. The advantages of the proposed method in recognition accuracy and testing time are all verified by extensive experiments with ablation studies.