Secondary Radar Signal Processing Based on Deep Residual Separable Neural Network

Abstract

In addition to the target echo, the signal environment of the secondary surveillance radar (SSR) also includes environmental echo and noise interference, which greatly affect the correct decoding of the response signal. Nowadays, the processing of secondary radar response signals basically uses traditional signal processing methods, and there are few methods for denoising using deep learning neural networks. This paper proposes a secondary radar signal processing method based on the deep residual separable neural network (DRS-Net), which can effectively extract the deep features of the secondary radar signal and predict the original response signal. The core of the network is based on the deep separable convolutional neural network, and the deep residual connection structure can effectively learn the deep features of the signal. We conducted a lot of experiments and verifications using secondary radar response signals with different signal-to-noise ratio noise. The experimental results show that the method has high denoising performance in the normal radar operating environment and can accurately predict secondary radar response signals. When the signal-to-noise ratio is 5dB, the strict accuracy rate can reach 94%. When the signal-to-noise ratio is higher than 10dB, the strict accuracy rate has reached 99.95%.