Approach of Specific Communication Emitter Identification Combining Differential Variable Modal Decomposition and Global Feature Analysis

‍ ‍To solve the problem of inadequate decomposition between modal components in the Hilbert-Huang transform （HHT） based special emitter identification （SEI）， a new SEI method combined signal processing and deep learning was proposed. Firstly， the raw signal was differenced and the corresponding intrinsic modal components （IMF） were obtained by variational modal decomposition （VMD）. Then， the Hilbert spectrums were obtained by Hilbert transform of each IMF. Finally， for the sparsity of the Hilbert spectrum， this paper invoked the global context block and improved it to further extract the global subtle features from the Hilbert spectrum. The performance of the proposed method was tested using ORACLE public dataset， and the experimental results showed that the recognition rate of the proposed method is better than four existing methods of SEI based on Hilbert spectrum， which had low computational complexity and had more than 90% recognition rate at 5 dB signal to noise ratio （SNR）.