Wavelet-Based Noise Reduction in Power Analysis Attack

Side-channel power analysis attacks have been proven to be the most powerful attacks on implementations of cryptographic primitives. DPA and CPA are probably the most wide-spread practical attacks on numerous embedded cryptographic systems. Additive noise is a kind of typical power analysis resistant implementing technique. The success rate of the DPA and CPA attacks is significantly affected by the Signal-to-Noise Ratio (SNR) of the power traces. Hence, it is important to eliminate noise effectively and improve the SNR for power attacks to extract the secret key. In this paper, a new method is proposed with wavelet analysis to reduce noise effects, which aims to improve the performance of side-channel power attacks named as WPA (Wavelet-based Power Attack). The key is successfully recovered on commercially available contact-less smart card based on 3DES, which is widely used for security-sensitive applications. Experimental results show that WPA significantly improves the success rate for encryption key detection. Compared to the fourth-order cumulant noise reduction method for power attack, the required number of the power traces is 50% less.