An Efficient Correlation Power Analysis Attack Using Variational Mode Decomposition ☆

Side channel attacks (SCAs) is now a real threat to cryptographic devices and correlation power analysis (CPA) is the most powerful attack. So far, CPA attack usually exploits the leakage information from raw power consumption traces that collected from attack device. In real attack scenarios, these traces collected from measurement equipment are usually contaminated by noise resulting in a decreases of attack efficiency. In this paper, we propose a variant CPA attack that exploits the leakage information from intrinsic mode functions (IMFs) of the power traces. These IMFs are the results of variational mode decomposition (VMD) process on the raw power traces. This attack technique decreases the number of power trace for correctly recovering the secret key by approximately 13% in normal condition and 60% in noisy condition compared to traditional CPA attack. Experiments were performed on power traces collected from the Atmega8515 smart card run on the Sakura-G/W side channel evaluation board to verify the effectiveness of our method.