Research on the Grouping Method of Side-Channel Leakage Detection

Power analysis attack is a method to obtain the key of the cryptographic chip by analyzing the correlation between power consumption information leaked by the cryptographic chip during the computing process and the key. The efficiency of power analysis attack poses a serious threat to the software and hardware implementation of cryptographic algorithms. In order to detect whether a cryptographic chip has information leakage, it is necessary to assess it by using detection techniques. The t-test is a hypothesis test used in the field of statistics to test whether there is a significant difference in the means of two normally distributed populations with unknown variance, and is also a useful tool in side-channel information leakage assessment. In this paper, two grouping methods are proposed based on the characteristics of the AES algorithm to investigate the construction of two overall groups before the implementation of the Welch’s t-test. Experimental verification of the DPA contest V4 dataset shows that both grouping methods were effective in detecting a large number of leakage points on power traces, but the grouping method by AES first round S-box output Hamming weight has a higher proportion of both the number of leakage points and the high t-statistic distribution than the method of grouping by bit value.