An Automated Application-Independent Approach to Anomaly Detection in Wireless Sensor Networks

Authors

Abstract

As Wireless Sensor Networks (WSN) gain momentum in what concerns applications and deployment, monitoring is becoming crucial in order to guarantee that anomalies are promptly detected. Unfortunately, current WSN monitoring solutions have several limitations, such as being tailored for specific applications, requiring dedicated or specific hardware, consuming precious en-ergy and/or processing resources, or relying on manual or offline intervention. In this paper we propose an approach to anomaly detection in WSNs that ad-dresses these limitations. The approach is based on two very simple metrics, a logging tool, and a data-mining algorithm, thus leading to the following key characteristics: very low resource consumption, application independency, very good potential for multi-WSN monitoring, and automation and simplification of the detection process. The proposed approach was validated by implementation, which showed that it is quite effective in detecting several typical anomalies.

Subject

Wireless sensor networks

Conference

WWIC 2014 - 12th International Conference on Wired & Wireless Internet Communications, Paris, France, 26-28 May 2014., May 2014

PDF File