Outlier Preprocessing in Wireless Sensor Networks: A Two-Layered Ellipse Approach

Abstract

Sensor nodes in wireless sensor networks have limited energy resources and this hinders the dissemination of the gathered data to a central location. This stimulated our research to make use of the limited computational capabilities of these sensor nodes to build a normal model of the data gathered. Hence by having the normal model, anomalies can then be detected and forwarded to a central location. This process is done locally in the sensor nodes and hence reduces the power consumption used in transmitting all the data. Our algorithm is an enhanced version of the Data Capture Anomalies Detection Algorithm, which is used to compute a local model of the normal data in wireless sensor networks. In this paper the Data Capture Anomalies Detection is used to partition the data and then send all the data to a central server for data classification, building on the Data Capture Anomalies Detection method and in order to classify the partitioned data, our algorithm Two-Layered Data Capture Anomalies Detection sends anomalies (2%) as well as roughly (2% or 4%) of normal data for further data processing and classification purposes. Experimental results on synthetic data show that Two-Layered Data Capture Anomalies Detection is able to provide promising results.