Using the ANOVA F-Statistic to Rapidly Identify Near-Field Vulnerabilities of Cryptographic Modules
V Iyer, Ali E. Yılmaz · 2021
The analysis of variance (ANOVA) F-statistic is proposed as an indicator to accelerate the identification of nearfield vulnerabilities of cryptographic modules to electromagnetic side-channel analysis (EM SCA) attacks. It is hypothesized that optimal measurement configurations that require collecting the fewest measurements to disclosure have high F-values; i.e., in these configurations, the measured signals exhibit high variability when the encryption changes and low variability when the encryption is repeated. The concept is demonstrated for an EM SCA attack to disclose the secret key used in an open-source implementation of the Advanced Encryption Standard (AES). It is shown that the F-statistic reduces the search space to identify optimal measurement configurations 6× to 17× depending on the probe height and orientation.