Knowledge-based reasoning for network fault diagnosis

Because of network connection and the relations among services and network devices, a network fault usually happens with a series of related faults. These cases bring complications to find out the exact original fault. This paper proposes a knowledge-based fault analysis approach KRAN. We first present its model and then give a rule-based reasoning algorithm for fault reasoning. This paper also implements a prototype of KRAN in a broadband access network and uses the real fault cases to validate its efficiency. Our efforts show that KRAN has high accuracy in fault diagnosis.