Fault Diagnosis For WSNs Using Time Domain Features Of Sensing Data

Due to the limit of resources, dynamic network topology and uncertainty, Faults are inevitable in the running process of network. A single critical node failure or a particular region nodes fails will lead to a net work failure within a specified range around it. Which will result in network connectivity partition. It would greatly reduce the quality of service in wireless sensor network system and the function of wireless sensor networks will be weakened or fail. Therefore, timely and accurate troubleshooting of wireless sensor networks in order to ensure reliable and efficient operation of the network becomes obviously important. The existing diagnostic methods consume a lot of communication bandwidth and node resources, which lead to heavy burden of the resources-limited network. fault diagnosis for wireless sensor networks is difficult.This paper presents a diagnosis method used for fault detection and fault classification based on the time domain features of sensing data(TDSD). Firstly, the feature extraction and analysis of the sensing data are carried out using one-dimensional discrete Gabor transform, mining the characteristics of the fault. Then the data are diagnosed and classified with SOM neural network, finally the current network status and identify the fault cause are determined. The faults were detected and classified mainly through temperature, humidity and other sensory data combined with the voltage data.It combined the feature extraction and analysis of the sensing data with the one dimensional discrete Gabor transform algorithm with SOM neural network, based on a series of rules failure knowledge library, monitor the network performance.In order to further develop the efficiency and accuracy of the diagnostic algorithm,design a wireless sensor network data monitoring system, it can clearly show the nodes data trend, capable of reacting node failure data features. Judging the current network nodes statu, and finding the corresponding fault type.The system is simple, custom format,and strong versatility and practicality.The results show that, comparing with DSD algorithm, this method has fewer burdens in network communication, better diagnostic accuracy rate and classification results, etc., and it has a high diagnostic accuracy especially for both node fault and network fault. Our algorithm has better detection rates in large-scale WSNs diagnosis. Its detection rate is more than 97%.and the false alarm rate is less than 40% when the network nodes reached 160.