A Bayesian Network approach to diagnosing the root cause of failure from Trouble Tickets

Abstract

Telecommunications networks comprise elements of very different types that work together to provide services. Quite often, hardware failures are interrelated and it is hard for technicians specialized in specific hardware to find out these relationships. In this context, Bayesian Networks (BN) provide a good and flexible solution because they allow us to model the causal relationships between element failures and infer information from existing evidence. The goal is that network technicians can be informed of the real scope of failures and the probable existence of root problems, thus optimizing resources and reducing recovery time. Besides, with this approach a real element hierarchy can be built, allowing the discovery of hidden dependencies between elements. The outcome of this work has been the development of a rooting module attached to an incident management system (trouble ticketing system, TT).