DRAT :A Penetration Testing Framework for Drones

As the usage of drones increases, attack vectors to exploit the vulnerabilities increases as well; particularly in commercial off-the-shelf Wi-Fi based drones. Hence, drones must be carefully evaluated and selected before deployment in the field. Penetration testing is a way to assess the vulnerabilities of drones, but it may require multiple commands, files or scripts and tools to generate and store the results. Many of these existing techniques and tools are dependent on human control and intervention. In this paper, we propose a Drone Pen-testing tool, which has integrated resources to conduct, organize the penetration tests and store the results. The tool has 3 main operation modes-Admin mode, User Mode and Machine learning mode. In Machine learning mode, the tool passively collects the network traffic from Wi-Fi drone access points for 60 seconds. The collected network traffic (in a pcap file) is used to analyze the IEEE 802.11 b/g/n protocol stack to identify a specific target among the surrounding Wi-Fi drones. This feature helps the user to launch a targeted attack quickly for a particular type of drone when the surrounding has many active drones. The paper explains the features of the scalable and easy to use, GUI (Graphical User Interface) based framework including details of its machine learning mode.