Resource-aware provisioning strategies in translucent elastic optical networks

Abstract

Elastic optical networks (EONs) enable a more efficient use of spectrum resources than traditional wavelength-division multiplexing (WDM) networks, mainly by means of a flexible frequency grid and the use of bandwidth variable transponders (BVTs). In WDM networks, regeneration is avoided because it always produces a cost increase and it is only recommended when the transmission length cannot be met. Instead, in EONs, a trade-off exists between regeneration and spectrum use. Indeed, blocking probability can be widely reduced by incorporating regeneration (BVTs in a back-to-back configuration) to shorten the transmission lengths, therefore increasing spectrum efficiency. Provisioning algorithms that take into account this trade-off consider regeneration either as a cost to be minimized (full regeneration capacity) or as a resource to be smartly used (bounded number of predeployed transponders).In recent years, algorithms that minimize the resource use of either spectrum or BVT have been proposed. These algorithms, however, penalize the overall network performance in terms of blocking probability and cost, since they do not take into account the available resources. In this work, we propose two novel resource-aware algorithms, that take into account all available resources in the provisioning process. We report simulation results showing that using these resource-aware algorithms: (i) when full regeneration capacity is assumed, the regeneration cost can be reduced by more than a 30% compared to the opaque solution and by a 10% compared to the best known algorithms; (ii) for the bounded case, resource-aware strategies can reduce blocking probability by thousands of times compared to the transparent solution and by hundreds of times compared to the best known algorithms. Our results show that resource awareness helps to decrease the blocking probability while increasing the actual network capacity, thanks to a more efficient assignment of both spectrum and BVT resources.