The rapid growth of the Internet of Things (IoT) has led to an increased automation and interconnectivity of devices without requiring user intervention, thereby enhancing the quality of our lives. However, the security of IoT devices is a significant concern as they are vulnerable to cyber-attacks, which can cause severe damage if not detected and resolved in time. To address this challenge, this study proposes a novel approach using a combination of deep learning and three-level algorithms to detect attacks in IoT networks quickly and accurately. The Bot-IoT dataset is used to evaluate the proposed approach, and the results show significant improvements in detection performance compared to existing methods. The proposed approach can also be extended to enhance the security of other IoT applications, making it a promising contribution to the field of IoT security.

The Internet of Things (IoT) contributes to improving and automating the quality of our lives via devices and applications that progressively become more interconnected without user intervention in many areas such as smart homes, smart cities, smart transportation, and smart environment. However, IoT devices are vulnerable to cyberattacks. We cannot prevent all attacks, but they can be detected and resolved with the least damage. Moreover, they are connected for long periods of time without user intervention. Additionally, since they remain connected for long periods of time without user intervention, creative solutions must be devised to keep them safe, such as machine learning. The reach goal is to evaluate different machine learning algorithms to detect IoT network attacks quickly and effectively. The Bot-IoT dataset, which is derived from the original dataset, is used to evaluate various detection algorithms. Five different machine learning algorithms were tested on the two databases, and the results of the tests revealed high and accurate performance at all levels of the dataset.