A Wavelet-based Power Analysis Attack Against Random Delay Countermeasure.
Xiaofei Dong,Fan Zhang,Samiya Queshi,Yiran Zhang,Ziyuan Liang,Bolin Yang,Feng Gao
DOI: https://doi.org/10.1109/asianhost.2018.8607162
2018-01-01
Abstract:Random delay insertion is a simple yet rather effective technique to increase the difficulty for traditional power analysis. However as compared to the random masking technique, it is uncommonly used as a countermeasure considering the frequency analysis. In this paper, it is investigated that the frequency analysis may not work as efficiently as expected when facing to advanced random delay countermeasures. Hence, a novel attack is proposed which is in the wavelet domain. After preprocessing the wavelet coefficients of power traces with wavelet decomposition, the effects of multiple random delays can be removed. Two attack strategies are proposed to recover the secret key: either indirectly from the reconstructed power traces without random delays or directly from the processed wavelet coefficients. Our experimental results show that the wavelet-based power analysis attack can perform much better than those frequency-based ones, which is evaluated through several standard metrics to show the efficiency and robustness.