A Model-Based Approach for Outlier Detection in Sensor Networks

摘要

In this paper, we propose a model-based approach to detect outliers in sensor networks by exploring the spatial correlation among neighboring nodes. This research is motivated by the observation that sensors in close proximity normally present similar readings. We propose to employ Gaussian mixture modeling as a statistical means to build a probability density function for multivariate spatial neighborhood sensor readings. Outlying sensors can be reliably detected since they exhibit extremely low density values. Our extensive simulation evaluation validates the proposed model-based approach for outlier detection in sensor networks.