On Comparison of Countermeasures against Statistical Ineffective Fault Attacks

Statistical Ineffective Fault Attacks were proposed in CHES 2018 as a new powerful, yet practical, fault model that can be used to leak information from implementations that are protected against forgery, side-channel or traditional fault attacks. The attack leaks information about messages where the attack is ineffective. In other words, when the fault targeted by the attacker matches the internal value of the state, the fault goes unnoticed by traditional countermeasures, which give the attacker enough information about the internal state of the device. In May 2019, three potential countermeasures for this type of attack were proposed. These countermeasures have different goals and strategies, making each of them more convenient in a different scenario. In this paper, we characterize the strategies used in each of these countermeasures and study the cost of implementing each of them. We also study the security advantages and disadvantages of each of them and show some of the shortcomings of these countermeasures.