联合对角化算法在功耗分析去噪中的应用

The Application of JADE Denoising in Power Analysis

提出了一种应用联合对角化算法(Joint Approximative Diagonalization of Eigen-Matrics,JADE)提高功耗信息信噪比的方法。在采集了功耗数据的基础上,使用联合对角化算法对功耗数据和噪声进行分离,使信噪比提升了33.6d B。并使用快速固定点算法(Fast ICA)与之对比,在1000次实验条件下,统计出两种算法提升信噪比的平均值和串音误差值。结果表明,联合对角化算法比快速固定点算法提升的信噪比高3.7d B,而联合对角化算法串音误差也低于快速固定点算法,证实其在功耗分析去噪中更有优势。

This paper proposed a novel Joint Approximative Diagonalization of Eigen-Matrics （JADE） to improve the power information of the signal to noise ratio （SNR）. JADE was employed to separate power data and noise with the collected power data, and the SNR increased by 33.6dB. In this experiment, the performance of the proposed approach was compared with fast fixed point Independent Component Analysis （FastlCA）. The average value of increased SNR and crosstalk error were calculated with 1000 times of experiments. The experimental result indicated that the proposed approach yielded 3.7dB higher in SNR enhancement than FastICA, and the crosstalk error is also smaller than FastICA, which proved the advantage and effectiveness of the proposed approach in de-noising