Cellular Learning Automata-Based Virtual Network Embedding in Software-Defined Networks

Abstract

Software-defined networking (SDN) is a propitious technology for achieving network virtualization by decoupling the control and data planes of a network. SDN hypervisor supports multiple virtual SDN-based networks logically isolated from each other. Each virtual SDN has its own controller and allocated resources over physical network. For achieving optimal resource allocation, there is a need of efficient virtual network embedding (VNE) approach in multidomain virtual SDN-based network. In this paper, we propose a self-adjusted, online, distributed virtual network mapping strategy based upon the idea of irregular cellular learning automata. We consider two aspects of the network during the execution of VNE in SDN—node and link mapping, and optimal placement of SDN controller. We evaluate the proposed scheme vSDN-CLA using Mininet. The simulation results show significant performance improvement in terms of throughput and end-to-end delay. Considering a substrate network of 100 nodes, we observed that the proposed scheme achieved 23.72 and 10.55% higher throughput, and 28.13 and 42% lesser end-to-end delay compared to that in two benchmark schemes DM-vSDN and CO-vSDN, respectively.