Side Channel Attack on SM4 Algorithm with Ensemble Method

Abstract

Smart IC card may undergo attacks that beyond cryptanalysis. Side Chanel Attack(SCA) gathers information leaked from implementations of smart IC card to find the secret key. Recent Publications had shown that it is possible to apply machine learning algorithms on SCA analysis. This article shows how to apply the ensemble method, one of the state of art methods, on SM4 algorithm. Ensemble method uses multiple algorithms to get better result. We apply SVM (support vector machine) and k-NN (k nearest neighbors) algorithms, compared with traditional template attack (TA) method. We investigated the impact of changing the distance function on the accuracy of the k-NN classifier. We also make a brief analysis on why does the ensemble method work on this problem.