A weighted statistical analysis of DPA attack on an ASIC AES implementation

Research on Differential Power Analysis (DPA) is becoming more and more popular nowadays. Against different hardware, DPA attack will lead to different results, e.g. attack on FPGA is easy but on ASIC is relatively difficult. However, how could we draw the conclusion that we finish a successful attack? How to build the connection between result data and successful rate is a meaningful topic. Same data with different statistical processing method will lead to different results. In this paper, we focus on DPA attack against an ASIC implementation. We analyzed two current statistical methods, pointed out the limitations. Then, we proposed a weighted statistical analysis method. According to number of traces and repetitions, we adjust the weight value. Finally, we improved the accuracy and efficiency of DPA attack on ASIC by reducing the number of traces used in attack.