Location constrained virtual optical network embedding in space-division multiplexing elastic optical networks

Virtual optical network embedding (VONE) involves two processes, node mapping, and link mapping. Node mapping decides which data center to place the virtual nodes (VNs). Link mapping decides how to route virtual links between placed VNs. In location constrained VONE (LC-VONE), each VN can be mapped only to some substrate nodes. In this paper, we study the problem of LC-VONE in space-division multiplexing elastic optical networks (SDM-EONs). Aiming at load balancing, we formulate the first integer linear programming (ILP) to solve the static LC-VONE problem, where the set of VON requests is given. To achieve scalability, an integrated framework is devised to jointly conduct node and link mappings for each VON request. Three cost metrics are designed for finding the least-cost VON solution. Combining them with the integrated framework, three LC-VONE algorithms are proposed: minimum spectrum consumption (MSC), bottleneck link avoidance (BLA), and lowest traffic load (LTL). By applying them to solve both static and dynamic LC-VONE problems, we found that LTL always outperforms MSC and BLA.