An Efficient Embedding Algorithm for Energy Multi-Domain Virtual Network Embedding

Abstract

Network virtualization (NV) is one important attribute of the next generation network (e.g, upcoming 5G virtualized networking). Virtual network embedding (VNE) is about the resource allocation problem in NV. Multiple VNE algorithms have been proposed over the past decade. Prior algorithms focus on accepting more virtual networks (VNs) in one shared substrate network (SN) in order to maximize the embedding revenue. As energy cost is approaching half of the SN operating cost, it is crucial for infrastructure providers to minimize energy cost. In addition, in real networking environment, it is vital to embed each VN among multiple SNs, geographically distributed in different regions. To address both issues, we firstly propose the formal VNE problem model and VNE energy cost model. Then an energy efficient heuristic algorithm (EEMD-VNE) is detailed, aiming at maximizing the net profit. Meanwhile, the EEMD-VNE is able to embed each requested VN among multiple SNs in polynomial time. Numerical simulation results reveal that the EEMD-VNE significantly reduces the energy cost by approximately 8.5% over the existing energy related algorithms for accommodating the same amount of VNs.