5G network slices embedding with sharable virtual network functions
Chengli Mei, Jiayi Liu, Jinyan Li, Lei Zhang, Menghan Shao
Abstract
Network slicing (NS) is recognized as a key technology for the 5G mobile network in enabling the network to support multiple diversified vertical markets over a shared physical infrastructure with efficiency and flexibility. A 5G NS instance is composed of a set of virtual network function (VNF) instances to form the end-to-end (E2E) virtual network for the slice to operate independently. The deployment of a NS is a typical virtual network embedding (VNE) problem. We consider a scenario in which VNF instances can be shared across multiple slices to further enhance the utilization ratio of the underlying physical resources. For NSs with sharable VNF instances, the deployment of the slice instances is essentially the embedding of multiple virtual networks coupled by the VNFs shared among slices. Hence, we formulate this sharable-VNFs-based multiple coupled VNE problem (SVM-VNE) through an integer linear program (ILP) formulation, and design a back-tracking coordinated virtual network mapping algorithm. Simulation results demonstrate that VNF-sharing can enhance the slice acceptance ratio with the same physical network, which represents higher physical resource utilization. Moreover, our approach achieves higher acceptance ratio by comparing to a baseline algorithm.