VNE-RFD: Virtual network embedding with resource fragmentation consideration

Abstract

Virtual Network Embedding (VNE) is one of the critical techniques in network virtualization. In this paper, we attempt to improve the performance of VNE in terms of acceptance ratio of Virtual Network (VN) requests by considering resource fragmentation in a substrate network. Different from existing work, we involve a new metric called Resource Fragmentation Degree (RFD) to quantitatively measure the status of resource fragmentation at substrate nodes and links. The definition of RFD is derived from the conception of connectivity in graph theory. Actually, RFD of a node (or a link) is only determined by residual resources at its neighbor nodes and adjacent links. Thus it can be derived efficiently. For a node (or a link), maximum RFD is reached when either all of its neighbor nodes or all of its adjacent links are running out of resources. In this case, resources at the node (or the link) are entirely fragmented. With consideration of RFD, we then formulate the problem of VNE as a mixed integer program. The optimization objective includes minimizing fragmented resources indicated by RFD in the substrate network. Finally, an online algorithm called VNE-RFD is proposed to solve the problem. Simulation results show that VNE-RFD can effectively reduce resource fragmentation and thus accept more VN requests compared with some existing algorithms.