Correlation Power Analysis for SM4 Based on ISOMAP

In Correlation Power Analysis (CPA), the problem is that the attacking rate becomes slow when there are many power sampling points. We introduced a method of preprocessing power data based on ISOMAP. Before performing CPA, this method uses the ISOMAP algorithm to reduce the dimension of power matrix. Under the guarantee of the intrinsic dimension of the power consumption matrix, the computational complexity of the CPA is reduced. At the same time, after the dimension is reduced, the data correlation is enhanced, which improves the attack efficiency of CPA.