Fault diagnosis on wireless sensor network using the neighborhood kernel density estimation

Wireless sensor network (WSN) has become one of the most important technologies because of its reliable remote monitoring ability. As sensors are often deployed at remote and/or hazardous environments, it is important to be able to perform faulty sensor nodes self-diagnosing. In this paper, we formulate WSN faulty nodes identification as a pattern classification problem. This paper uses semi-supervised method for faulty sensor nodes classification. To enhance the learning performance, we also introduce a label propagation mechanism which is based on local kernel density estimation. The basic concept of the method is to estimate the posterior probability of a scene that belongs to normal or different faulty modes. In this paper, we implemented a software platform to study WSN under different number of sensor nodes and faulty conditions. Our experimental results show the proposed semi-supervised method is highly effective. Thorough comparative analyses with other state-of-art semi-supervised learning methods were included. The obtained results confirmed that our proposed algorithm can deliver improved classification performance for WSN.