Spectrum allocation using multiparameter optimization in elastic optical networks

The Min Slot-Continuity Capacity Loss (MSCL) algorithm is listed in the literature as a powerful algorithm to solve the spectrum assignment (SA) problem in elastic optical networks (EONs). The MSCL presents, in its structure, characteristics that allow it to consider several unique network parameters that impact the network performance during the decision of the spectral allocation, such as: the evaluation of the impact of a spectral allocation in both the candidate route and in its interfering routes and the evaluation of the impact of a spectral allocation on the network's capacity to successfully allocate future demands. However, other interesting features can be still added to the MSCL. Thus, in this paper we propose a framework that enables a systematic design of heuristics SA algorithms based on the MSCL principles. The main building blocks of the framework are the MSCL principles, the inclusion of new metrics and how to associate them, a generic function expanded in a series of functions and the utilization of a methaheuristic optimization algorithm. We apply the proposed framework to design two new SA heuristics named as MPAO-Wj and MPAO-WLj. These heuristics use the particle swarm optimization as optimization engine. We carried out computational simulations to assess the performance of the proposed algorithms and we compared them against five other benchmark heuristics from the literature. The simulation results indicate that, in the considered scenarios, the proposed algorithms are able to achieve superior network performance (in terms of blocking probability) than the investigated benchmark algorithms.

Automated summary

The MSCL algorithm is a powerful algorithm that solves the spectrum assignment problem in elastic optical networks. This paper proposes a framework based on the MSCL principles to design SA algorithms and applies it to two new SA heuristics. The simulation results show that the proposed algorithms achieve superior network performance compared to benchmark algorithms.