SmartIDS: A Comparative Study of Intelligent Intrusion Detection Systems for Internet of Things

Abstract

AbstractTraditional intrusion detection systems (IDSs) are not scalable and efficient in detecting intrusions in IoT systems; hence, protecting them against cyber-attacks. The need to secure the Internet of Things (IoT) platforms deployable at a large scale to build smart systems gave rise to a new class of intelligent and scalable IDSs. Intelligent IDSs, which employ machine-learning (ML) or deep learning (DL) methods, have shown promising results in detecting intrusions with high accuracy and better detection rates than traditional IDSs that suffer from scalability, low detection rates, and inefficiency issues. This paper presents a comparative analysis of a selected set of intelligent IDSs using the Microsoft Azure ML Studio (AML-S) platform and datasets containing malicious and benign IoT network traffic.KeywordsIntrusion detectionAnomaly detectionInternet of ThingsClassificationMachine learningDeep learning