Double-Key Recovery Based Correlation Power Analysis

Abstract

Side Channel Analysis (SCA) provides a powerful tool for skilled adversaries to reduce the effort to determine an unknown key by the principle of 'divide and conquer', especially for the implementation of symmetric cryptographic algorithm. Among the SCA techniques, Correlation Power Analysis (CPA) is one of the most famous methods since it is easy to carry out and the efficiency is usually high. The traditional CPA divide the whole key into several subkeys according to the cryptographic algorithm and analyze the subkeys one by one. This method has a problem that the power traces can not be ensured to be most efficiently utilized since that for every analysis the power traces are used once. In this paper, we creatively propose a new 'divide and conquer' scheme. We divide the whole key into several double-keys, for every double-key we bring forward an innovative concept of chaining of power traces. This method can effectively improve the efficiency of CPA, and it is useful when the power traces are not enough. We prove the availability of our method by theory knowledge. Besides, we perform a series of practical experiments on both the public data and the power data that was captured by ourselves to evaluate the key-recovery efficiency of double-key based CPA. The experimental results verify that, in the same attack scenario, the key-recovery efficiency of double-key based CPA is higher than that of traditional CPA.