Abstract
Firmware fuzzing has shown to be a useful tool to discover vulnerabilities in embedded devices. However, applying fuzzing effectively to real firmware — such as that used in wireless medical sensors and patient monitoring equipment — requires expertise and careful configuration. In this paper we present Aurora-Fuzzware, a workflow and configuration tool. It enables automatic configuration for the Fuzzware framework for Nordic nRF52-family of devices, and a structured approach for setting up a fuzzing process. We validated our approach through three 24 hour fuzzing experiments with different configurations, and through a questionnaire to one of our industry partners. The results show that (i) the configuration tool successfully generates working configurations for nRF52 binaries developed with the NSDK 15.3, (ii) workflow-guided optimizations identified through Aurora-Fuzzware significantly improved coverage, and (iii) while the configuration tool is relatively user-friendly, there remains clear room for improvement. To bridge the gap between academic prototypes and practical industrial use, future work should emphasize improving the usability and user experience of fuzzing frameworks.