Black Hole Traffic Anomaly Detections in Wireless Sensor Network

With the flourish of Internet of Things, the security issues in wireless sensor network (WSN), especially traffic anomaly detections, have attracted researchers' attentions. As a distributed wireless network, WSN is vulnerable to many attacks. In this research, the authors investigate the traffic anomaly detections of a well-known attack, black hole attack, in WSNs. With limited computation capacity, sensor nodes are unable to perform sophisticated detection techniques. Therefore, the authors propose a profile based monitoring approach with a restricted feature set to supervise the network traffic. The proposed profile based monitoring approach contains two components, feature selection and anomaly detection. In order to complement the limited computing capacity of a sensor node, feature selection component will extract features with high contribution or high relevance for future monitoring. The anomaly detection component monitors the selected features and alarms the administrator when an anomaly is detected. Two types of combination are proposed, graphic and non-graphic based models. The graphic based approach seems to surpass the non-graphic based approach, but the graphic based approach takes much longer time to select the important features than non-graphic based approach.