Collaborative Attributes and Resources for Single-Stage Virtual Network Mapping in Network Virtualization

Haotong Cao, Hongbo Zhu, and Longxiang Yang

10.1109/JCN.2019.000045

Abstract : Virtual network embedding (VNE) is the virtualized node and link resources allocation problem in network virtualization environment, aiming at achieving the simultaneous optimal node and link mapping assignment per VN. Currently, a large number of mapping algorithms for VNE exist in the academia. Existing algorithms mostly focus on mapping each virtual network (VN) in two ordered stages: First virtual node mapping stage and second virtual link mapping stage, leading to non-optimal VN mapping assignment. Though multiple one-stage algorithms exist, adopting either optimization theory or graph theory, they usually involve much more VN assignment calculation time. Hence, these one-stage algorithms are not suitable for dynamic network scenario. Not to mention practical VNE application. Based on the background and our gained research results, we propose another heuristic mapping algorithm VNE-ARS, completing all nodes and links per VN in one single embedding stage. Each VN embedding assignment can be completed in polynomial time, using our VNE-ARS. Main network attributes and resources are collaborated, serving as the mapping criterion of our VNE-ARS. In order to highlight our VNE-ARS strength, we conduct the evaluation experiments. We compare our VNE-ARS algorithm against existing mapping algorithms.  

Index terms : Heuristic algorithm, mapping algorithm, network attributes, one single stage, two ordered stages, virtual network embedding.