Network Anomaly Detection in Wireless Sensor Networks: A Review

Abstract

Wireless sensor networks function as one of the enablers for the large-scale deployment of Internet of Things in various applications, including critical infrastructure. However, the open communications environment of wireless systems, immature technologies and the inherent limitations of sensor nodes make wireless sensor networks an attractive target to malicious activities. The main contributions of this review include describing the true nature of wireless sensor networks through their characteristics and security threats as well as reflecting them to network anomaly detection by surveying recent studies in the field. The potential and feasibility of graph-based deep learning for detecting anomalies in these networks are also explored. Finally, some remarks on modelling anomaly detection methods, using appropriate datasets for validation purposes and interpreting complex machine learning models are given.