A Study on Machine Learning Approaches for Outlier Detection in Wireless Sensor Network

Abstract

Wireless Sensor Network (WSN) is an important research area nowadays. Wireless Sensor Network is deployed in hostile environment consisting of hundreds to thousands of nodes. They can be deployed for various mission-critical applications, such as health care, military monitoring as well as civilian applications. There are various security issues in these networks. One of such issue is outlier detection. In outlier detection, data obtained by some of the nodes whose behavior is different from the data of other nodes are spotted in the group of data. But identification of such nodes is a little difficult. In this paper, machine learning based methods for outlier detection are discussed among which the Bayesian Network looks advantageous over other methods. Bayesian classification algorithm can be used for calculating the conditional dependency of the available nodes in WSN. This method can also calculate the missing data value.