Fuzzing Drone Control System Configurations Based on Quality-Diversity Enhanced Genetic Algorithm

As drones are becoming widely used in various fields, drone security is a growing challenge nowadays. Drone control systems use various configuration parameters to control their positions and attitudes. If these parameters are misconfigured, drones will fall into abnormal flight states, such as trajectory deviation and crash to the ground. Existing works mainly focus on system memory errors which lead to obvious system failure but don’t apply to drone flight state anomalies. This paper focuses on abnormal drone flight states caused by configuration parameter errors. We propose a novel state-guided fuzzing system called APFuzzer, which searches for incorrect configuration parameter values that would trigger abnormal flight states. To enhance the capability of searching for multiple optimal solutions, we design a quality-diversity enhanced genetic algorithm (QDGA) to mutate configurations to search for incorrect configuration parameter values and consider the effects of environmental factors and flight missions on the flight states. We evaluated APFuzzer on the drone control system ArduPilot and successfully searched 3389 incorrect configuration parameter values and triggered all predefined five abnormal flight states. In addition, APFuzzer automatically analyzed the fuzzing results and found five software bugs related to configurations.