Influence of transients on end-to-end availability for a meshed network

The disruption of services must be kept at a minimum in telecommunications networks so that the consequences are not too severe and their durations are as short as possible. Maintenance policies often rely on the steady-state availabilities of each element of the system, and focus on the system’s weak links. The end-to-end (or two-terminal) availability – a standard performance index – of a meshed network has long been studied, but mainly for small systems, and assuming constant values for the availability of each element. When taken into account, the time-dependent contributions of links and nodes to the system unavailability were computed using exponential failure and repair distributions. In this work we revisit the meshed network first proposed by Walter, Esch, and Limbourg (ESREL 2008), and compute the end-to-end availability between two nodes, where the individual contributions of links and nodes are kept. This allows the ranking of links and nodes, using well-known performance indices (Birnbaum, Risk Reduction Worth, etc.). We can thus determine the elements that should receive due attention in maintenance and resilience studies. However, as the steady-state availability may not always be a lower bound to the transient availability in the case of non-exponential failure and repair distributions, we have studied the influence of such configurations on the time-dependent behaviours of all the aforementioned quantities. We then discuss the influence of uncertainty in the availability values, and compare the results obtained for the all-terminal reliability, another often-used performance criterion of networks.

---

Opracowanie rekordu ze środków MEiN, umowa nr SONP/SP/546092/2022 w ramach programu "Społeczna odpowiedzialność nauki" - moduł: Popularyzacja nauki i promocja sportu (2022-2023).