Energy-efficient, EDFA lifetime-aware network planning along with virtualized elastic regenerator placement for IP-over-EON

TLDR: This paper focuses on energy-efficient network planning (including traffic provisioning) along with optimal placement of virtualized elastic regenerators (VERs) for IP-over-elastic optical networks based on a static traffic profile, using a mixed integer linear programming-based optimization model.

Abstract: In this paper, we focus on energy-efficient network planning (including traffic provisioning) along with optimal placement of virtualized elastic regenerators (VERs) for IP-over-elastic optical networks based on a static traffic profile, using a mixed integer linear programming-based optimization model. The proposed model judiciously exploits flexibility of IP core routers, sliceable bandwidth variable transponders (SBVTs) and VERs to accommodate the traffic demands with the minimum power consumption (PC). Optical layer traffic grooming allows to simultaneously originate/terminate multiple lightpaths of different capacities, data slots and maximum transparent reach by a single SBVT. The proposed model also allows to use all functionalities of VER, such as simultaneous regeneration, distance-adaptive transmission option selection, frequency slot merging to be used concurrently for the given static traffic profile. In addition, the proposed model includes lifetime awareness of Erbium-doped fiber amplifier (EDFA) to reduce EDFA failure and associated repairing cost in long run. Using the proposed model, we enhance the average EDFA lifetime by restricting average EDFA occupancy, represented by the ratio of the (average) number of lightpaths being amplified in an EDFA and the maximum possible number of lightpaths that can be amplified in it, even though in the process, the overall PC in network may increase, with reference to the scenario with no EDFA occupancy restriction. The variation in PC and average EDFA lifetime for different permissible (user-defined) average EDFA occupancy are studied. We exhaustively study performance of the model under different network conditions.