An Effective Tool for Traces Preprocessing in Side-Channel Analysis: Ridge Energies Extraction From Synchrosqueezing Wavelet Transform
Yuanzhen Wang, Hongxin Zhang, Shaofei Sun, Yunfei Ma, Xiong Wei, Zhi H. I. Sun, Md Sabbir Hosen · IEEE Internet of Things Journal · 2025
In recent years, deep learning methods have become prevalent in the field of side-channel analysis (SCA), leading to a decline in research on nonprofiled attacks and their preprocessing techniques. However, while deep-learning-based SCA methods can reduce the requirements for the quality of leakage signals, they do not fully resolve the need for preprocessing in SCA. Additionally, there is a lack of universal, nontrained preprocessing methods for side-channel leakage under various types of interference. In this article, a new method for preprocessing side channel leakage using synchrosqueezing wavelet transform and ridge energy extraction is proposed to deal with complex scenarios. This method involves two steps: first, applying synchrosqueezing theory to concentrate signal energy in the wavelet domain, and second, utilizing a dynamic path optimization algorithm to locate and extract ridge energy for subsequent attacks. Our proposed solution extracts key information features from the traces, effectively addressing trace distortion caused by varying levels of noise and random disturbances. We validate the effectiveness of our analysis and solution through extensive experiments on four public databases and two self-collected datasets containing power and electromagnetic leakage (EM) from both hardware and software encryption implementations. We compare our experimental results from four indicators: 1) success rate; 2) guessing entropy; 3) signal to noise ratio; and 4) normalized interclass varianc. Besides, we demonstrate the advantages of our method by comparing Measurements to Disclosure and on public datasets. Our method demonstrates significant potential for preprocessing traces in nonprofiled SCA.