Research On Power Analysis Attack For Block Cipher Algorithm Of Smart Card

At present, cryptographic chips which represented by smart cards are used more and more widely. Smart cards made an important position for the feature of portability and security in the finance, medical, transportation and authentication and so on. The appearance of smart cards also greatly improved the level of modernization of people’s life and work. But, there are a variety of attacks and malicious intrusions which aimed at smart cards after its popularity. With the appearance of Power Analysis Attacks(PAA), the cryptographic chip’s security was at a risk, and because the cryptographic chip’s security had a close relationship with the national security and social stability, so it is very necessary to have a research on smart card with power analysis attacks.Based on the study of power analysis attack on smart cards, we mainly completed the following three works.(1) We have established the experimental platform of our attacks which was based on the software of Inspector SCA for the company of Riscure.(2) We proposed the most appropriate method of attacks to complete the power analysis attacks for classical algorithms of DES and AES(For DES, we used the principle of Mean Difference and chose the output of S boxes to conduct a DPA attack; for AES, we chose the output of Add Round Key as the specific location of DPA attacks). Meanwhile, we designed the reasonable process to complete corresponding attacks which based on the experimental platform that we established at first, which showed that the effectiveness of power analysis and the threat to smart cards. Selecting the appropriate method of attacks can effectively achieve the result of attack and reduce the error peaks and improve the efficiency of attacks in the power analysis attack.(3) Aiming at the phenomenon “ghost peaks” in the result of DPA, we put forward a new method called IDPA, which was based on a standard DPA attack, and it was an improvement of DPA attack. IDPA made a comparison between the peak distribution of real attack and the predicted peak distributions of assuming no noise attack. Then, we used the Euclidean similarity to match the peak model between real measurement and predicted measurement. We proved that the IDPA indeed improved the phenomenon “ghost peaks” and its attacking efficiency with the experiment of hardware test and software simulation.