Enhanced Correlation Power Analysis by Biasing Power Traces

• Published in: Information Security Vol. 9866; pp. 59 - 72
• Main Authors: Ou, Changhai, Wang, Zhu, Sun, Degang, Zhou, Xinping, Ai, Juan, Pang, Na
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 2016; Springer International Publishing
• Series: Lecture Notes in Computer Science
• Subjects: Biasing power trace; Computer security; CPA; Power trace; Side channel; Success rate
• ISBN: 3319458701; 9783319458700
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-319-45871-7_5

Biasing power traces with high Signal to Noise Ratio (SNR) proposed by K. Yongdae et al. can significantly improve the efficiency of the CPA. But it is still a problem to be solved that how to efficiently select power traces with high SNR. Through the analysis of the statistical characteristics of power traces, we propose three methods to better solve this problem in this paper. We bias power traces by using the Minkowski distance (i.e. Euclidean distance or Manhattan distance) between each power trace and mean power trace. Biasing power traces can also be carried out by using probability density function values of power consumption of interesting points, or even directly using power consumption of interesting points. Our schemes can blindly select power traces with high SNR in a high probability. The efficiency of the CPA by using the three of our methods is significantly improved. Thus, our schemes are more effective compared to the one proposed by K. Yongdae et al.