Internet of Things (IoT) devices are becoming a part of our daily life; from health monitors to critical infrastructure, they are used everywhere. This makes them ideal targets for malicious actors to exploit for nefarious purposes. Recent attacks like the Mirai botnet are just examples in which default credentials were used to exploit thousands of devices. This raises major concerns about IoT device security. In this work, we aimed to investigate security of IoT devices through performing automatic penetration test on IoT devices. A penetration test is a way of detecting security problems, but manually testing billions of IoT devices is infeasible. This work has therefore examined autonomous penetration testing on IoT devices. In recent studies, automated attack execution models were developed for modeling automated attacks in cyber ranges. We have (1) investigated how such models can be applied for performing autonomous IoT penetration testing. Furthermore, we have (2) investigated if some well known and severe Wi-Fi related vulnerabilities still exist in IoT devices. Through a case study, we have shown that the such models can be used to model and design autonomous penetration testing agents for IoT devices. In addition, we have demonstrated that well-known vulnerabilities are present in deployed and currently sold products used in IoT devices, and that they can be both autonomously revealed through our developed system.
Keywords: IoT; cybersecurity; penetration testing.