Virtual network mapping algorithms based on multi-level reliability sorting in multiple domain networks

Abstract

Multi-domain network is a large-scale underlying network, which is connected to each other through links but independent of each other, and network virtualization methods are used for virtual network mapping in a multi-domain network environment. How to better allocate physical resources to virtual networks and ensure that the average failure rate of the network is minimized is the focus of the current network virtualization research field. Most of the existing research on multi-domain network environment aims to improve the resource utilization of the network, ignoring the generation of network fragmentation and the failure rate of nodes and links in the network, and does not consider the differences between each network domain. Based on the current research, we propose a virtual network mapping algorithm based on multilevel reliable ordering in multi-domain networks (MRS-VNE). First, we adopt a hierarchical collaborative multi-domain network mapping framework, then propose a multi-level reliable ordering of network domains and physical nodes, prioritize intra-domain virtual network mapping, and finally, adopt heuristic inter-domain virtual network mapping for VNRs that fail intra-domain mapping. Simulation results show that the algorithm can improve the request acceptance rate and reduce the overall average failure rate.