VNEavXT: Offline Virtual Network Embedding Model Considering Crosstalk-Avoided Approach in Spectrally-Spatially Elastic Optical Networks

Abstract

Infrastructure as a service and network virtualization facilitates the network provider/operator to lease out the resources among different stakeholders, hence promoting the sharing of network resources among stakeholders. At the same time, spectrally-spatially elastic optical networks (SS-EONs) have emerged as a prominent solution to release pressures out of the backhaul network by overcoming the physical barrier and expanding it in the spatial dimension. The efficient mapping of virtual optical network (VON) requests is challenging in SS-EONs because of the additional constraints. For the first time, this work proposes a crosstalk (XT)-avoided model for routing, spectrum, core, and mode allocation (RSCMA), named VNEavXT, to improve resource utilization by avoiding inter-core and inter-mode XT. VNEavXT allocates the network resources fulfilling the spectrum continuity, spectrum contiguity, core continuity, mode continuity, and XT-avoided constraints. An integer linear programming (ILP) problem is formulated for VNEavXT. We prove that the decision version of VNEavXT is NP-complete. Two heuristic algorithms are introduced for large instances where the ILP approach becomes intractable. Finally, their performances are evaluated by considering different performance metrics. We observe that the heuristic based on ILP embeds more requests than the heuristic based on rank and a benchmark heuristic designed for static (off-line) virtual optical network mapping, but in terms of computation time, the heuristic based on rank is faster than the heuristic based on ILP and the benchmark.