A Highly Effective DPA Attack Method Based on Genetic Algorithm

• 作者: Shuaiwei Zhang Weidong Zhong Xiaoyuan Yang Yujuan Sun
• 来源期刊: 计算机、材料和连续体(英文)
• 学科: 工学
• 关键词: DPA EFFICIENCY noise GENETIC algorithm FITNESS function novel model
• 年，卷（期）: 2018,（8）
• 页码范围: 325-338
• 页数: 14页
• 分类号: TP3

摘要

As one of the typical method for side channel attack,DPA has become a serious trouble for the security of encryption algorithm implementation.The potential capability of DPA attack induces researchers making a lot of efforts in this area,which significantly improved the attack efficiency of DPA.However,most of these efforts were made based on the hypothesis that the gathered power consumption data from the target device were stable and low noise.If large deviation happens in part of the power consumption data sample,the efficiency of DPA attack will be reduced rapidly.In this work,a highly efficient method for DPA attack is proposed with the inspiration of genetic algorithm.Based on the designed fitness function,power consumption data that is stable and less noisy will be selected and the noisy ones will be eliminated.In this way,not only improves the robustness and efficiency of DPA attack,but also reduces the number of samples needed.With experiments on block cipher algorithms of DES and SM4,10%and 12.5%of the number of power consumption curves have been reduced in average with the proposed DPAG algorithm compared to original DPA attack respectively.The high efficiency and correctness of the proposed algorithm and novel model are proved by experiments.