Centralized KNN anomaly detector for WSN

This work suggests to detect abnormalities from spatial distribution of data using a numerical outlier data detector in a wireless sensor network (WSN). The detector is able to find anomalous from one or many events by using KNN technique and Euclidian distance. WSN uses a Low Energy Adaptive Clustering Hierarchy protocol (LEACH), where we compute the good impact of detection on energy. From this, a mean time to failure is computed. The evaluation is also with detection rates metrics in order to appreciate the detection accuracy and quality of data.