One-Sided Countermeasures for Side-Channel Attacks Can Backfire

Abstract

Side-channel attacks are currently one of the most powerful attacks against implementations of cryptographic algorithms. They exploit the correlation between the physical measurements (power consumption, electromagnetic emissions, timing) taken at different points during the computation and the secret key. Some of the existing countermeasures offer a protection against one specific type of side channel only. We show that it can be a bad practice which can make exploitation of other side-channels easier. First, we perform a power analysis attack on an FPGA implementation of the Advanced Encryption Standard (AES) which is not protected against side-channel attacks and estimate the number of power traces required to extract its secret key. Then, we repeat the attack on AES implementations which are protected against fault injections by hardware redundancy and show that they can be broken with three times less power traces than the unprotected AES. We also demonstrate that the problem cannot be solved by complementing the duplicated module, as previously proposed. Our results show that there is a need for increasing knowledge about side-channel attacks and designing stronger countermeasures.