An outlier detection algorithm in wireless sensor network based on clustering

This paper presents a novel wireless sensor network outlier detection algorithm based on clustering (ODC). Firstly, the ODC algorithm defines time slot for data sampling. After that, ODC get reasonable clusters for all time slots with nodes as attributes. After the clustering process, it gets the maximum cluster and the minimum cluster. It assigns all time slots according to the distances to the two cluster centers. Summarizing the distances between sequential time slots alternatively labeled by different cluster in some time period, it gets the length of feature period T when it appears regular cycle. Finally, ODC finds latent outliers by detecting cluster labels of time slots. Experimental results on real public wireless sensor data sets are provided to illustrate the efficiency and the robustness of the proposed algorithm.