Research of Wireless Network Fault Diagnosis Based on Bayesian Networks

In the field of wireless network fault diagnosis, the relationship between the phenomenon and reason of fault is complicated and non-linear. So it adds a great deal of difficulty to wireless network optimizers when they are dealing with network problems. In response to this problem, in this paper, taken the CDMA network as an example, the method of wireless network fault diagnosis based on Bayesian Network is discussed. Two diagnosis models, Causation Bayesian Network model and Naive Bayesian Network model, are established and applied to experiment. They are testified to be precise and reliable and the result is used to evaluate the advantage and disadvantage of them. Besides, for the incompleteness of actual performance data, three mature incomplete-data-set learning methods, Monte-Carlo method, Gaussian algorithm and EM algorithm, are applied, whose function and shortcoming are explained.