Vulcan: a state-aware fuzzing tool for wear OS ecosystem

This demo abstract introduces Vulcan, a fuzz testing tool for evaluating the robustness of wearable device by injecting intra-device and inter-device communication messages. Vulcan first builds a state-model of a wearable app by offline training then steers the app to a target state for injecting mutated messages. The target state of the app typically runs a high number of concurrent processes. By testing a set of 100 popular Wear OS apps, Vulcan was able to trigger 45 unique crashes and 18 system reboots. These system reboots are triggered by a fuzzing user-level app and we present a mitigation strategy to prevent it.