SCD-NS: Service Customized Deployment of Network Slicing

Abstract

The fifth generation networks (5G) can support a variety of vertical industries, such as autonomous driving, telemedicine and industrial automation. Network Slicing (NS) is a promising technology for 5G that can provide customized end-to-end network services for multi-tenants. To provide services to tenants according to different service requirements, we model delay and reliability of NS. We propose a method for service-customized deployment of network slicing (SCD-NS), which allocates resources according to different types of network slices. When deploying a network slice, we adopt a two-stage deployment method, in which first stage is virtual node deployment and second stage is virtual link deployment. Since the problem of network slices deployment is usually NP-Hard, we encode the virtual node deployment solution as a chromosome and use genetic algorithms to solve this problem. In the stage of virtual node deployment, we obtain the initial solutions through minimizing the average number of hops between nodes. In the stage of virtual link deployment, we provide adaptive deployment based on the types of network slices. The simulation results show that SCD-NS realizes better service customization and higher network utilization compared with current algorithm.