Deep Learning for Anomaly Detection in IoT Time Series

Abstract

With the widespread adoption of the Internet of Things (IoT), time-series data is being generated in massive quantities across various industries. In IoT systems, detecting anomalies in time-series data is crucial for ensuring reliability, security, and efficiency. This chapter provides a comprehensive overview of anomaly detection techniques in IoT time-series data, categorising anomalies into three main types: point anomalies, collective anomalies, and contextual anomalies. Firstly, it discusses the significance of anomaly detection in IoT, highlighting its importance in diverse applications such as smart grid, network, and financial sectors. Subsequently, it surveys commonly used anomaly detection methods, including statistical approaches, machine learning algorithms, and deep learning models, outlining their principles, advantages, and limitations. Furthermore, challenges and future directions in IoT anomaly detection are discussed, addressing issues such as data heterogeneity, scalability, interpretability, and real-time processing. This review serves as a valuable resource for researchers, practitioners, and stakeholders interested in understanding and deploying anomaly detection techniques for IoT time-series data.