A Novel Intrusion Detection Framework for Internet of Things Based on Machine Learning Techniques

As a new paradigm of communication technology, the Internet of Things (IoT) profoundly transforms the world and substantially influences human society. Due to its inherent weaknesses and unfledged trait, IoT is an attractive target for hackers, making the defense of IoT systems an urgent necessity. Since machine learning (ML) technologies have rapidly evolved in recent years, their application in Intrusion Detection Systems (IDS) for enhancing IoT security has shown a promising prospect and has obtained considerable interest from research communities. In this research, we propose a novel IoT IDS framework based on machine learning techniques. The proposed framework is assessed in terms of six metrics by applying 10 classifiers on the IoTID20 dataset. The evaluation results demonstrate the effectiveness of the best-performing classifier and its superior performance to other existing models in previous works.