RTL Flow for the Power Side-Channel Resilience Assessment of a Post-Quantum SHA-3 Accelerator

Despite Post-Quantum Cryptography (PQC) algorithms exhibiting high security against computational attacks from both quantum and classical computers, they may still be vulnerable to physical attacks due to hardware/software implementation vulnerabilities. Therefore, resilience assessment of PQC designs against physical attacks, such as power Side-Channel Analysis (SCA) attacks, is crucial to provide designers with useful hints to improve the security of their PQC designs. In this paper, we propose an RTL flow for the resilience assessment against power SCA attacks of an RTL design that implements a hardware accelerator for the Keccak-f[1600], a SHA-3-based PQC hash function. The proposed RTL flow consists of EDA tools for power traces collection and a model that considers the resolution, the sampling frequency and the error of the adversary's measurements. For the resilience assessment against power SCA attacks, the guessing entropy metric is obtained by performing Deep Learning (DL) SCA on power traces of SHA-3-based algorithms. We present resilience results obtained by the proposed flow, when applied to the Keccak-1[1600] RTL design, and we identify characteristics of attack scenarios, such as adversary's measurements resolution, sampling rate and error, making PQC designs more vulnerable and prone to power SCA attacks.