Time-frequency Analysis and Convolutional Neural Network Based Fuze Jamming Signal Recognition

Abstract

Fuze jamming signal recognition plays a critical role in the battlefield environment. To improve the performance of fuze jamming signals detection, we propose a fuze jamming signal detector based on time-frequency analysis (TFA) and convolutional neural network (CNN), called TFA-CNN, in this paper. The detailed recognition process of the proposed TFA- CNN detector is provided, where the short-time Fourier trans- form (STFT) is first employed to convert the original jammed fuze signals into the time-frequency images and then the TFA- CNN detector is built to train the recognition model. Simulation results verify that the TFA-CNN detector outperforms the typical existing recognition detectors, such as LeNet, time-frequency images and convolutional neural network (TFI-CNN) and deep neural network (DNN), in the detection performance with a slightly higher time complexity. Specially, the average recognition accuracy of the proposed detector achieves 99.8% even at a low signal-to-interference-plus-noise ratio (SINR).