Breaking LPA-resistant cryptographic circuits with principal component analysis

In this paper, a novel hardware attack based on principal component analysis (PCA) is proposed to break a leakage power analysis (LPA)-resistant cryptographic circuit (CC) efficiently. Although the added false keys which are used for masking the secret key of the LPA-resistant CC are secure and effective against regular LPA attacks, they may be precisely modeled by eigenvalues and eigenvectors under PCA. After performing the proposed PCA on the LPA-resistant CC, all the added false keys can be removed to expose the corresponding secret key. As shown in the result, only 2000 number of plaintexts are sufficient to crack an LPA-resistant CC by utilizing the proposed PCA-assisted LPA attacks.