One class outlier detection method in wireless sensor networks: Comparative study

Abstract

Recent advances in communication technology have enable the emergency of new types of wireless networks: Wireless Sensor Networks (WSN). It consists of a huge number of tiny and low cost devices with sensing and communication capabilities. They are emerging recently as a key solution to monitor remote environments and concern a wide range of applications from the environmental and military surveillance to home automation. However, these models are not suitable for the energy constrained WSNs because they assumed the whole data is available in a central location for further analysis. In this paper, we present a comparative study between Centralised and Distributed One-class Outliers Detection Classifier (COODC & DOODC) based on Mahalanobis Kernel used for outlier detection in wireless sensor networks (WSNs). For this case, the task amounts to create a useful model based on KPCA to recognize data as normal or outliers. Recently, Kernel Principal component analysis (KPCA) has used for nonlinear case which can extract higher order statistics. On account of the attractive capability, KPCA-based methods have been extensively investigated, and have showed excellent performance. Within this setting, we propose Kernel Principal Component Analysis based Mahalanobis kernel as a new outlier detection method using Mahalanobis distance to implicitly calculate the mapping of the data points in the feature space so that we can separate outlier points from normal pattern of data distribution. The use of Distributed One-class Outliers Detection Classifier based on Mahalanobis Kernel on real word data obtained from Intel Berkeley are reported showing that the proposed method performs better in finding outliers in wireless sensor networks when compared to the Centralised One-class Outliers Detection Classifier (COODC).