Software Countermeasures against DVFS fault Attack for AES

Most modern processors utilize Dynamic Voltage and Frequency Scaling (DVFS) technology to dynamically adjust voltage and frequency according to the workloads of processors to reduce power consumption. However, DVFS can be exploited by attackers to introduce transient hardware faults. Combined with Differential Fault Analysis (DFA), the secure execution environment of processors is threatened. In order to minimize the threat of DVFS fault attack, we analyze its principle and focus on countermeasures for AES applications. We provide a detailed analysis of three software countermeasures, namely redundancy-based methods, parity code, and infective computation. Among them, the temporal redundant method with random latency between executions defends against high-order attacks with 34.18% timing overhead, while the infective computation-based method achieves the best security with nearly 124.35% timing overhead.