Feature extraction of side channel signal based on EMD and difference spectrum of singular value

Abstract

In order to extract feature information of side channel signal from the strong noise,the approach of signal feature extraction was proposed on account of empirical mode decomposition (EMD) and difference spectrum of singular value.Firstly,using EMD to decompose the original side channel signal,the correlation coefficient between all the intrinsic mode functions (IMF) and the original signal were calculated to find the component of biggest simi-lar characteristic.Then,to make a singular value decomposition of IMF component and corresponding difference spectrum of singular values was obtained.Finally,according to the difference spectrum,the filtered and recon-structed signal of IMF component was obtained,and the feature information of component was further extracted.The actual examples show that the method can extract characteristic information of side channel signal validly and improve the SNR and success rate of attack of signal successfully.

Keywords

• side channel singnal
• empirical mode decomposition
• difference spectrum of singular value
• feature ex-traction
• attack success rate