基于滑动平均的能量分析攻击研究与实现

Research and Implementation of Power Analysis Based on Moving Average

在能量分析攻击中,为了提高攻击效率,减少噪声的影响,越来越多的预处理方法被使用,取得了很多显著的效果。该文以在ATmega16上运行的AES-128算法作为攻击目标,对采集的原始能量曲线进行滑动平均滤波,再通过相关能量分析(Correlation Power Analysis,CPA)攻击确定滑动平均滤波的最优参数。通过与原始数据和经过Hanning窗低通滤波器滤波后的数据对比,可以看出经过滑动平均滤波处理后,使用正确密钥所得的CPA相关系数较原始数据或低通滤波处理后所得系数大,而错误密钥所得相关系数小。经过滑动平均处理后,不仅可以发现AES-128的10轮加密过程,而且经过差分能量分析攻击所得的尖峰较原始的更为明显。实验结果表明经过滑动平均滤波预处理后,能量分析攻击的效率可明显提高。

In order to improve the efficiency of attack and reduce the influence of noise on power analysis, a growing number of preprocessing methods are discussed and numerous remarkable results are reported. The AES-128 algorithm running on the ATmegal6 is taken as the target in this paper. The original energy curves are moved average and the optimal parameter of moving average filter is determined by Correlation Power Analysis （CPA） subsequently. The experimental results demonstrate that compared with the original data and the data after Hanning window low-pass filter, the correlation coefficient obtained by the correct key with the using of moving average filter is evidently promoted, while the correlation coefficient obtained by the incorrect key is decreased. With the moving average approaching, the process of the ten encryption of AES-128 can be discovered obviously. The peak of Differential Power Analysis （DPA）obtained by the data using moving average is more obvious than that obtained by the original data. Numerical results show that the moving average approaching can improve the efficiency of power analysis evidently.