Machine Learning Approach for Anomaly Detection in Wireless Sensor Data

Wireless sensor nodes can experience faults during deployment either due to its hardware malfunctioning or software failure or even harsh environmental factors and battery failure. This results into presence of anomalies in their time-series collected data. So, these anomalies demand for reliable detection strategies to support in long term and/or in large scale WSN deployments. These data of physical variables are transmitted continuously to a repository for further processing of information as data stream. This paper presents a novel and distributed machine learning approach towards different anomalies detection based on incorporating the combined properties of wavelet and support vector machine (SVM). The time-series filtered data are passed through mother wavelets and several statistical features are extracted. Then features are classified using SVM to detect anomalies as short fault (SF) and noise fault (NF). The results obtained indicate that the proposed approach has excellent performance in fault detection and its classification of WS data.