Advances in ML-Based Anomaly Detection for the IoT

Abstract

The Internet of Things drives many activities in our modern world. Through its heterogeneity and connectivity to the Internet, it provides an attractive and big attack surface. Anomaly detection is a central tool for making IoT systems more secure. Since 2017, machine learning is successfully used for anomaly detection. This work gives an overview on the evolution of using machine learning for anomaly detection including the most active research groups, and the most attractive venues. In addition, it discusses the advantages and disadvantages of the available methods based on their use in literature.