Open AccessSpecific emitter identification of radar based on one dimensional convolution neural network
Author(s) -
Yao Xiao,
Xi Wei
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1550/3/032114
Subject(s) - convolutional neural network , computer science , common emitter , radar , convolution (computer science) , artificial intelligence , fast fourier transform , artificial neural network , signal (programming language) , pattern recognition (psychology) , feature extraction , identification (biology) , frequency domain , time domain , feature (linguistics) , electronic engineering , telecommunications , algorithm , engineering , computer vision , linguistics , philosophy , botany , biology , programming language
Specific Emitter Identification (SEI) of radar has become a focus point of research and difficulty in the field of electronic reconnaissance, and the application of Automatic Dependent Surveillance Broadcast (ADS-B) signal in Identification Friend or Foe (IFF) system is beginning to raise increasing concern. At present, the research of deep neural network in the field of target recognition mostly focuses on the modulation pattern recognition of signal, but there are few applications for the specific emitter identification. In order to overcome the shortcoming that the effect of the traditional method is not excellent, a radar emitter target recognition method based on one-dimensional(1D) convolutional neural network is proposed for ADS-B signal in this paper. Taking the sequence data of ADS-B pulse signals as the training and testing samples of emitter target identification, the frequency domain features of ADS-B signal are obtained by Fast Fourier Transform (FFT), and the target identification is achieved by combining the convolution neural network with the fine feature extraction. The result of simulation demonstrates that the specific emitter identification based on frequency domain and one-dimensional(1D) convolutional neural network has an excellent recognition performance, which effectively resolves the shortcomings of the traditional method of feature extraction and low classification accuracy.