Fast Fault Localization for Large-scale IP-based Communication Systems Using Bayesian Networks

In large-scale IP-based communication systems, faults are often caused by multiple factors of uncertainty, so how to locate the root causes of these faultsquickly and precisely plays a key role in network management. This paper presents a fault localization technique based on Bayesian probabilistic inference. Equippedwith enhanced ability to handle negative and positivesymptoms, our method can diagnose multiple simultaneous faults and be resilient to noise in the symptom information. Moreover, it achieves good accuracy and precisionwith low cost of time. In the simulation study, we compareour approach with a prior fault localization technique IHU.Results show that our method offers better accuracy andprecision compared to IHU in noisy environment at the expense of less time especially in large-scale communicationsystems.