Evaluating Code Emulating Upsets on Microcontroller Devices

Fault attacks pose a significant threat to the functionality and reliability of embedded systems, particularly microcontrollers that are widely used in critical applications such as consumer electronics, medical devices, industrial systems, and recently machine learning models. Given the nature of fault injection attacks, the cost for setting up and running fault injection campaigns can be inaccurate, time consuming and expensive. Because of this, alternative methods such as software, GDB based and RTL model fault injection have been developed. This paper compares these fault injection methods and proposes a high-level, standalone method of injecting faults into the ARM Cortex M0+ microcontroller with high accuracy and customization, while exploring the vulnerability of various algorithms.