Migration of elastic optical networks to the C + L-bands subject to a partial upgrade of the number of erbium-doped fiber amplifiers

Space division multiplexing (SDM) and band division multiplexing (BDM) are considered promising technologies to increase the capacity of optical transport networks. The progressive shortage of available dark fibers and the immaturity of multicore and multimode fibers for multichannel transmission induce network operators to postpone the process of capacity enhancement through SDM. Therefore, capacity increase revolves around BDM by lighting up at least the L-band of the already installed optical fiber infrastructure, which is a practical solution in the short to middle term. However, L-band activation requires the upgrade of network components such as erbium-doped fiber amplifiers (EDFAs). To manage the imposed cost while leveraging the L-band, a network can be partially rather than fully migrated in a single step by upgrading just a subset of the fibers and thus a subset of EDFAs to operate in the C + L-bands. In this paper, the focus is set on determining which fibers in the network should be upgraded to exploit the L-band, subject to a constraint on the maximum number of EDFAs to be upgraded, and analyzing its impact on network performance when facing dynamic traffic in terms of the blocking ratio. To this end, three heuristic algorithms, each pursuing a different objective, and two of them based on an integer linear programming (ILP) formulation, are proposed for the network planning to identify which fibers to upgrade. Simulation results demonstrate that, thanks to the use of these heuristics, the upgrade of a partial set of links to the C + L line system is a viable solution for network operators to circumvent the huge cost associated with migrating the full network. For instance, we demonstrate that a strategic partial upgrade using the proposed methods, subject to upgrading a maximum of 60% of the EDFAs, can significantly boost the supported traffic load in the examined topologies, ranging from 175% to 322%, when compared to the non-upgraded network.