Network Slicing in Software-Defined Networks for Resource Optimization

Abstract

The Internet users' heterogeneous requirements pose a challenge to the service provider for catering to the network infrastructure needs to provide the different services. The partitioning of the shared physical network is proposed using a network slicing mechanism to meet this challenge. The classification of the services is carried out based on the attributes in the flow features. The flow features are extracted in real-time for traffic classification. The traffic classification is implemented using supervised learning models for different network services. The combination of traffic classification and network slicing is proposed for the dynamic allocation of resources. In this paper, the testbed to demonstrate the network slicing concept based on the bandwidth requirement is presented using FlowVisor and the OpenFlow protocol in software-defined networks. The intelligence is embedded into networks by integrating decision making using a supervised learning model for classification followed by network slicing for optimal resource allocation.