An adaptive singular value decomposition-based method to enhance correlation electromagnetic analysis

Abstract

Electromagnetic analysis in side channel attack exploits the information of electromagnetic radiation that leaks from the cryptographic devices when they are running. It's no-table because of its efficiency and easiness to perform. Correlation electromagnetic analysis (CEMA) is of the most effective means in electromagnetic analysis. However, the efficiency of traditional CEMA is limited by some insignificant' electromagnetic traces. It is necessary to select the helpful subset of the electromagnetic traces for analysis rather than using the whole electromagnetic traces set to improve the efficiency. In this paper, we first give an proposition about the CEMA and prove it by mathematical theory. This proposition illustrates the feasibility of selecting electromagnetic traces. Then we propose a method that is based on Singular Value Decomposition to select electromagnetic traces. This method is adaptive and doesn't need any external parameter. Besides, this method is useful for analyzing both unprotected implementation and masked implementation. We carry out the practical experiments by our SVD-CEMA and CEMA in the same scenario. The experimental results verify that the key-recovery efficiency of our method is higher than CEMA.