Optimizing Application-Oriented Bandwidth Allocation And Request Scheduling In Elastic Optical Networks

Recently, elastic optical networking has been one of the most important research focus in the optical networking field. Even though the flexible gird of EONs has many advantages, it also brings new challenges. Due to the fact that the sub-carriers that are for different applications, and have different modulation formats and bandwidth gran-ularities are coexistent in EONs, the real-time control and management of spectrum re-sources becomes complicated in the optical layer, more spectrum allocation constraints should be considered, and the effect of physical impairment cannot be ignored. Hence, it is crucial for EONs to realize the real-time control and management of spectrum re-sources in the optical layer, satisfying power constraint as well as quality of transmis-sion. Routing and spectrum allocation (RSA), or routing, modulation level and spectrum allocation (RMSA) is one of the key problem to be solved.Compared with the routing and wavelength assignment (RWA) problem in dense wavelength-division multiplexing (DWDM) networks, the RSA/RMSA problem in EONs becomes more complicated, due to their unique spectrum allocation mechanism, i.e., the constraints of spectrum contiguity, spectrum non-overlapping, and spectrum continuity should be considered in the spectrum allocation procedure, so that existing RWA so-lutions cannot be directly applied to the RSA/RMSA problem. Previously, researchers have worked on the RSA/RMSA problem in EONs, and proposed a variety of static/dynamic RSA/RMSA schemes. However, most of the proposed RSA/RMSA schemes are only for the applications that demand immediate reservation services. It’s known that, in reality, there are heterogenous applications that have diverse QoS parameters in the networks.Hence, in this work, we focus on how to optimize the bandwidth allocation and request scheduling for different applications in EONs. More specifically, we first in-troduce the bandwidth allocation and request scheduling problem in EONs, and gener-ally classify network applications according to their different QoS parameters on delay sensitivity and bandwidth awareness, into three types of requests, which are immedi-ate reservation, advance reservation, and malleable reservation requests, respectively. Then, we propose efficient algorithms for each of them, and simulation results verify the effectiveness of the proposed algorithms on reducing request blocking probability and improving spectrum utilization. Finally, we introduce the software-defined elastic optical networks (SD-EONs), and demonstrate an implementation of advance reserva-tion in a control plane test-bed for SD-EONs, in order to provide suggestions on how to apply the proposed algorithms in this work to the real network scenarios, and thus make this work valuable for both theoretical investigation and practical applications.