The model-based service fault diagnosis with probability analysis

With the development of web service, it´s important to localize the fault activity and explain the faulty reason. In this paper, we propose a model-based diagnosis method of web service fault. In our method, we firstly built service model using Petri nets, then rank the diagnosed activities by computing the faulty probability of all activities, finally use the defined diagnosis rules to find out the faulty activity in the given diagnosis sequence. Our method not only improves the diagnostic accuracy, but also raises the diagnostic efficiency. Our experimental results show our method is effective and outperforms the existing model-based method.