• High Power Laser and Particle Beams
  • Vol. 36, Issue 4, 043018 (2024)
Yunfeng Kou1, Fei Dai2, Zhiguo Zhao3, Jianming Lü1, and Xie Ma1
Author Affiliations
  • 1Chengdu Xinxinshenfeng Electronics Co, Ltd, Chengdu 611731, China
  • 2Beihang University, Beijing 100083, China
  • 3China Electronics Technology Cyber Security Co., Ltd, Chengdu 610041, China
  • show less
    DOI: 10.11884/HPLPB202436.230186 Cite this Article
    Yunfeng Kou, Fei Dai, Zhiguo Zhao, Jianming Lü, Xie Ma. Leakage signal classification and recognition method based on fusion features[J]. High Power Laser and Particle Beams, 2024, 36(4): 043018 Copy Citation Text show less
    Algorithm flowchart
    Fig. 1. Algorithm flowchart
    Wavelet feature projection
    Fig. 2. Wavelet feature projection
    Hilbert characteristic projection map
    Fig. 3. Hilbert characteristic projection map
    Bispectral feature projection maps
    Fig. 4. Bispectral feature projection maps
    Confusion matrix of prediction results based on fusion features when the signal-to-noise ratio is 0 dB
    Fig. 5. Confusion matrix of prediction results based on fusion features when the signal-to-noise ratio is 0 dB
    Confusion matrix of prediction results based on wavelet features at different signal-to-noise ratios
    Fig. 6. Confusion matrix of prediction results based on wavelet features at different signal-to-noise ratios
    Confusion matrix of prediction results based on HHT features at different signal-to-noise ratios
    Fig. 7. Confusion matrix of prediction results based on HHT features at different signal-to-noise ratios
    No.signal typetotal sampling points of each WAV filenumber of samples intercepted by each WAV filetotal number of samples taken
    1clock leak signal11264000,10035200, 7168000 ,7782400563,501,358,3891811
    2laptop touchpad leak signal12247040,15589376, 17924096,21274624612,779,896,10633350
    3environmental radio emissions signal17981440,22003712,25976832, 25075712 ,15302656899,1100,1298,1253,7655315
    4screen display signal21553152,34586624,267223041077,1729,13364142
    5unknown radiation source signal15728640,17661952, 26402816,16826368786,883,1320,8413830
    Table 1. Five types of leakage sources
    No.signal typewavelet feature mapHHT feature mapbispectral feature map
    1clock leak signal
    2laptop touchpad leak signal
    3environmental radio emissions signal
    4screen display signal
    5unknown radiation source signal
    Table 2. Five types of leak source characteristics
    No.signal typebalanced dataset sample sizeunbalanced dataset sample size
    training settest settraining settest set
    1clock leak signal14403601449362
    2laptop touchpad leak signal14403602680670
    3environmental radio emissions signal144036042521063
    4screen display signal14403603313829
    5unknown radiation source signal14403603064766
    Table 3. Number of sample data sets of five types of leakage sources
    No.SNR/dBfusion feature map/%wavelet feature map/%HHT feature map/%bispectral feature map/%
    1099.895.895.2100
    2310098.497.8100
    3510093.698.8100
    4710093.099.8100
    Table 4. Prediction accuracy of different feature maps under different signal-to-noise ratios
    Yunfeng Kou, Fei Dai, Zhiguo Zhao, Jianming Lü, Xie Ma. Leakage signal classification and recognition method based on fusion features[J]. High Power Laser and Particle Beams, 2024, 36(4): 043018
    Download Citation