A Bayesian Approach for Automated Troubleshooting for UMTS Networks

Troubleshooting (TS) in UMTS networks are basic management tasks required to guarantee efficient usage of the network infrastructure. This paper presents a methodology for automating TS tasks based on Bayesian networks (BN). In a first learning phase, data relating symptoms and alarms to faults in the network is extracted and used to create the TS model. In a second phase, symptoms and alarms are used to identify faults with the highest probabilities. A TS case study using a dynamic system simulator illustrates the effectiveness of the proposed approach