Research On Fault Diagnosis Of WSN Based On Information Fusion

The wireless sensor networks (WSN) is one of the four innovative and hightechnology in the21th century. It is an integration of sensor technology, wirelesscommunication technique, distributed information processing technology and micro electromechanical system. It is the head of the top ten emerging technologies which have theprofound influence on human life and it also has a big application value in each domain.But the probability of sensor nodes’ failure is much higher than other systems due to manyinevitably factors and the complex and harsh environment. The fault message will reducethe whole networks’ service quality or even cause the whole network paralyzed. Therefore,the study for fault diagnosis of nodes in WSN is very necessary.The information fusion is a fashionable research method at present. It diagnoses faultinformation after integrating multi-source fault information so that it can improve theaccuracy greatly. In this paper, collect various kinds of fault symptoms and integrate themfirst and then diagnose the fault nodes through Rough Set theory (RS) and Least SquareSupport Vector Machine (LS-SVM). Firstly, the rough set theory is used to reduce theattributes of sampling data in order to select the decision-making attributes for constitutinga new simply dataset. Then using the new simply dataset to train the LS-SVM. And finallyclassify failure modes of WSN through the trained LS-SVM model.The algorithm in this paper makes full use of the RS and the LS-SVM. It not onlyfully develops the reduction ability of the RS and the excellent group ability of theLS-SVM, but also makes up the shortage of the RS which has a weak ability of antijamming capability and be sensitive of noise. At the same time, it overcomes the shortageof the LS-SVM model which can’t recognize the valid data effectively. It can reduce thedimension of input space so that it improves the diagnosis efficiency. Finally, contrastingdiagnosis method based on the RSLS-SVM model with the RS and the LS-SVMrespectively. The result shows the effectiveness of the algorithm in fault diagnosis of WSN,and also shows that it can improve the efficiency and accuracy of fault diagnosis.