Fault Diagnosis for High-Level Applications Based on Dynamic Bayesian Network

In order to improve the quality of Internet service, it is important to quickly and accurately diagnose the root fault from the observed symptoms and knowledge. Because of the dynamical changes in service system which are caused by many factors such as dynamic routing and link congestion, the dependence between the observed symptoms and the root faults becomes more complex and uncertain, especially in noisy environment. Therefore, the performance of fault localization based on static Bayesian network (BN) degrades. This paper establishes a fault diagnosis technique based on dynamic Bayesian network (DBN),which can deal with the system dynamics and noise. Moreover, our algorithm has taken several measures to reduce the algorithm complexity in order to run efficiently in large-scale networks. We implement simulation and compare our algorithm with our former algorithm based on BN (ITFD) in accuracy, efficiency and time. The results show that our algorithm can be effectively used to diagnose the root fault in high-level applications.