An Intrusion Detection System for the Internet of Things Using Machine Learning Models

The growing applications of the Internet of Things (IoT) technologies have led to enormous security problems. Since Intrusion Detection Systems (IDSs) based on Machine Learning (ML) techniques have a powerful capability in detecting intrusion, especially zero-day attacks, this domain has gained great attention from researchers. In this paper, we use a newly published IoT dataset, the TON_IoT network dataset, to build an IoT IDS. After comparing ten ML algorithms and comprehensive consideration, we choose the XGBoost method as the ultimate classifier. Experiments show that our model can achieve an MCC of 99.84% and 99.17% in binary and multiclass classifications, respectively.