Improved wavelet transform for noise reduction in power analysis attacks

Abstract

In side channel attacks (SCA), noise has been a hot topic for affecting the quality of obtained observations. In this paper, we propose a kind of improved wavelet transform denoising method based on singular spectral analysis (SSA) and detrended fluctuation analysis (DFA). Principal signal component in SSA can be selected by DFA adaptively, and residual part can be denoised by wavelet transform to retrieve important information. The method of superposition between signal component and denoised residual part improves the denoising efficiency of original wavelet transform. In order to verify the usefulness of the proposed method, we choose the correlation power analysis (CPA) to attack hard implementation of AES by using wavelet transform and the proposed method for preprocessing. Results show that the proposed method improve the success rate whilst decrease the necessary number of power consumption traces significantly. And the proposed method outperforms wavelet transform in noise elimination.