On the use of error correcting and detecting codes in secured circuits

When designing circuits for security-related applications, the robustness depends on its capability to globally resist to various types of attacks, based either on the observation of the device (e.g. differential power analysis) or its perturbation (e.g. injecting faults with a laser). To deal with perturbations, two strategies are developed: detecting the attack or maintaining the proper functioning of the device. This paper discusses characteristics of these two types of approaches when using information redundancy.