摘要

Detection of anomalous node in distributed wireless sensor networks is extremely important for powerful inference and network reliability. In this paper, we propose a powerful linear statistical model for estimating the state values of the sensor nodes longitudinally, and the estimated state values are used for detecting the anomalous nodes. Our proposed approach is powerful because it considers the effect of the nearest neighbors on the current state values and then detects the anomalous nodes based on the estimated state values. Our method can estimate the missing state values of the sensor nodes, which are kept in sleep mode for energy conservation. We also propose an alternative Bayesian model that is computationally faster for state estimation and anomaly detection. The effectiveness of the proposed model is investigated through extensive simulation studies, and the usefulness of our algorithm is numerically assessed. The performance of the proposed approach is compared to that of the traditional approaches through simulation studies. The proposed model can be effectively used in security surveillance, pattern recognition, habitat monitoring, etc.

  • 出版日期2017-6