Outlier Detection Techniques in WSNs A Survey

Abstract

 WSNs carries great potential in research and has colossal impact on emerging field of data communication networks, artificial intelligence and information technology. They are used in several critical applications like remote patient monitoring system, military surveillance, radiation monitoring, smart agriculture, fire detection etc., where decision making process has high dependency on the quality of data acquired from WSN. However, the raw data collected from sensors is highly vulnerable to noise while unusual real-time events can be easily subject to malicious attacks. To resolve this, the node central to the system must implement outlier detection algorithms for smooth system progress. Data classification becomes mandatory to prevent illogical behavior of the system so that techniques like data mining and machine learning can play key role in WSN improvements. It is necessary to examine data for outliers before analyzing and making decisions, thus outlier detection provides a shielding mechanism for WSNs against erroneous data which leads to fallacious operations. In this paper, we present a review on basic outlier detection techniques in WSNs .The survey can help to evaluate different techniques and can offer suggestions for future research.