Developing a fault prediction model for wired copper networks under precipitation

摘要

Telecommunication companies who face challenges of aging infrastructure need to balance the cost of maintenance with that of providing their services within a service level guarantee. For Telstra, the largest telecommunication company in Australia, this balance is achieved by adopting a passive approach to handle the faults that occur in the network. Rather than actively preventing faults, technicians are assigned to fix faults in a timely manner. However, to achieve an efficient and timely technician assignment, a prediction model is needed to advise planners of the potential number of faults in the network. From statistical analysis, we have developed a fault prediction model by investigating 29 months of data of faults. Our prediction model shows that rain has a significant impact on the number of faults in many areas across Australia, which can be the result of the aging infrastructure.