A Generic And Real-Time Internet Of Things Attack Realization And Detection Testbed

Official URL: https://risc01.sabanciuniv.edu/record=b3205816

The rapidly evolving Internet of Things (IoT) systems, which transform ordinary objects into interconnected smart devices, offer numerous advantages such as improved data collection and real-time analysis, leading to enhanced efficiency and resource utilization across various sectors. However, with the expansion of IoT, the security threats and vulnerabilities of these devices have also increased. Ensuring robust security measures, including secure data transfer and detecting potential attackers become crucial for maintaining data privacy, integrity, and availability. In this thesis, we underline the significance of IoT testbeds and the role of intrusion detection systems in mitigating these risks and vulnerabilities. We construct a low-cost IoT testbed in order to emulate generic IoT devices from a security perspective. By the use of collected network data, which include both benign and malicious traffic, a machine-learning-based real-time attack detection system is developed to detect and classify 17 types of cyber attacks. With F1 score of 0.9504 and recall value of 0.9524, the results show that the proposed real-time intrusion detection system can detect various cyberattacks on IoT systems.