Acta Metallurgica Sinica(English letters) ›› 2008, Vol. 15 ›› Issue (4): 101-106.doi:
• Artificial Intelligence • 上一篇 下一篇
韩煜,邹雪城,刘政林,陈毅成
HAN Yu, ZOU Xue-cheng, LIU Zheng-lin, CHEN Yi-cheng
摘要：
This article examines vulnerabilities to power analysis attacks between software and hardware implementations of cryptographic algorithms. Representative platforms including an Atmel 89S8252 8-bit processor and a 0.25 μm 1.8 v standard cell circuit are proposed to implement the advance encryption standard (AES). A simulation-based experimental environment is built to acquire power data, and single-bit differential power analysis (DPA), and multi-bit DPA and correlation power analysis (CPA) attacks are conducted on two implementations respectively. The experimental results show that the hardware implementation has less data-dependent power leakages to resist power attacks. Furthermore, an improved DPA approach is proposed. It adopts hamming distance of intermediate results as power model and arranges plaintext inputs to differentiate power traces to the maximal probability. Compared with the original power attacks, our improved DPA performs a successful attack on AES hardware implementations with acceptable power measurements and fewer computations.
