A Comparative Analysis of Machine Learning Deep Learning Approaches for Intrusion Detection in IoT/IIoT Networks

Abstract

The fast development of the Internet of Things and the Indus-trial Internet of Things has significantly increased the attack surface of current networks, where heterogeneous and resource-constrained devices work in ever-changing traffic conditions. Intrusion Detection Systems are an important security measure that is usually installed at gateways, edge nodes, or cloud plat-forms in order to track traffic and detect anomalies or known attack patterns. Although recent machine learning and deep learning techniques have such positive results are reported to be over 99% detection accuracies on benchmark datasets are commonly acquired in controlled experimental conditions and provide black box models that do not transparency is a severe constraint in safety-critical IoT/IIoT implementations where explainability and responsibility is necessary.