Machine Learning-based Anomaly Detection in ZigBee Networks

Abstract

With the development of information technology, IoT devices are spreading rapidly. ZigBee is one of the short-range wireless communication standards used in IoT devices and is expected to be used in smart homes and industrial control systems because of its low power consumption and low-cost operation despite its low communication speed. However, ZigBee can be subject to cyber-attacks because eavesdropping on packets and sending forged packets against wireless communication is easier than wired ones. In order to use ZigBee safely in smart home and industrial control systems, it is necessary to develop a method to detect cyber-attacks quickly. In this paper, we propose a machine learning-based anomaly detection system for Zigbee networks. We focus on characteristics of ZigBee communication and investigate a method to detect network anomalies and cyber attacks on ZigBee networks using machine learning. Furthermore, since we primarily put emphasis on practicality, our proposed system is simple and consists of widely used tools such as Wireshark. To evaluate the detection accuracy of our proposed system, we conduct some experiments. As a result, it is shown that our proposed system can detect attacks with high accuracy. In addition, we varied the features used in machine learning and discuss which feature has a high contribution to anomaly detection.