Outlier Detection Techniques and Cleaning of Data for Wireless Sensor Networks: A Survey
Vipnesh Jha, Om Veer Singh Yadav · 2012
Pattern recognition is the scientific discipline where the goal is the classification of objects into a number of categories or classes. Pattern recognition is an integral part in most sensing networks built for outlier detection. The significant deviations from the pattern of sensed data are considered as outliers in wireless sensor networks. These outliers include noise, errors, and malicious attack on the network. This affects the performance of the wireless sensor networks. Mostly the nature of sensor data is multivariate but it may be univariate also. Because of this, the traditional techniques are not directly applicable to wireless sensor networks. This contribution overviews existing outlier detection techniques developed for wireless sensor networks. It also presents a outlier detection technique framework to be used as a guideline to select a technique for outlier detection suitable for application based on the characteristics, such as, data type, outlier type and outlier degree.