Global Optimization of All-Optical Hybrid-Casting in Inter-Datacenter Elastic Optical Networks

Abstract

In inter-datacenter elastic optical networks (IDC-EONs), routing, modulation-level, and spectrum assignment (RMLSA) is one of the key measures to increase network capacity and flexibility of spectrum resource management. The coexistence of unicast, anycast, multicast, and manycast traffic in IDC-EONs forms a more complex network communication scheme called hybrid cast. However, the majority of the previous RMLSA approaches focused on the network communication scheme with single traffic type and optimized network performance on the specific network metric, which leads to less practicability and a limit to one-sided network performance improvement. To achieve better network performance, we investigate the hybrid-cast RMLSA (HC-RMLSA) problem with the objective of improving IDC-EON overall performance. Specifically, a criterion is defined to estimate the overall network performance by considering both the overall network service and operation effectiveness and the network cost. An ILP model is formulated to solve the HC-RMLSA problem under static network scenario. Then, we propose an independent optimization genetic algorithm with priority-based rationality adaption (IOGA-PRA) by coordinating a set of hybrid-cast requests simultaneously to solve the HC-RMLSA problem under both static and dynamic network scenarios. Numerical results show that the performance of IOGA-PRA is very close to the optimal solutions obtained by the ILP model on the network cost-effectiveness in static network scenario. The IOGA-PRA also achieves outstanding improvement of the overall network performance compared with benchmark algorithms in the dynamic network scenario.