An optimized cross correlation power attack of message blinding exponentiation algorithms

The message blinding method is the most efficient and secure countermeasure against first-order differential power analysis(DPA). Although cross correlation attacks(CCAs) were given for defeating message blinding methods, however searching for correlation points is difficult for noise, misalignment in practical environment. In this paper, we propose an optimized cross correlation power attack for message blinding exponentiation algorithms. The attack method can select the more correlative power points of share one operation in the modular multiplication by comparing variances between correlation coefficients. Further we demonstrate that the attack method is more efficient in experiments with hardware implementation of RSA on a crypto chip card. In addition to the proposed CCA method can recovery all 1024bits secret key and recognition rate increases to 100% even when the recorded signals are noisy.