An Intelligent Intrusion Detection System for the Internet of Things (IoT) Ecosystem

In this work, we propose a distributed intrusion detection system based on multi-agent and machine learning techniques specifically designed for the Internet of Things (IoT) ecosystem. The aim is to leverage multi-agent systems to promptly detect and respond to potential threats. To achieve this, we use the IOTID20 dataset, which provides a realistic overview of prominent IoT attacks, including DOS, MIRAI, and MITM attacks. We have selected and employed seven advanced machine learning algorithms to enhance intrusion detection effectiveness. Additionally, we propose a feature selection method based on a grey wolf optimizer to improve system performance. We achieved notable results, with a 100% success rate in binary detection and 91 % accuracy in subcategory detection. These results align with previous research, improvement. Moreover, we addressed IoT dataset limitations by implementing a majority voting system to ensure optimal outcomes.