MARViC: Multi-attribute Driven Resource Allocation For Virtual Network Embedding In Cloud Networks

Abstract

Network virtualization enables service providers to partition the physical substrate into multiple logically isolated virtual networks (VNs), each comprising virtual machines (VMs) interconnected via virtual links (VLs). Mapping virtual network requests (VNRs) onto the substrate network, known as virtual network embedding (VNE), is an NP-hard problem. Existing VNE approaches primarily optimize system-level metrics, such as resource utilization and the revenue-to-cost ratio, while often overlooking the topological dependencies among virtual and substrate nodes. To address this limitation, we propose MARViC, a Multi-Attribute driven Resource allocation framework for Virtual network embedding in Cloud networks. MARViC models VNE as a multi-criteria decision-making (MCDM) problem and jointly incorporates network and system-level attributes to capture structural dependencies and improve embedding decisions. Specifically, it employs entropy-based weighting to determine the relative importance of individual attributes and VIKOR-based ranking to prioritize both VMs and substrate servers. Using these rankings, MARViC performs greedy VM placement followed by shortest-path-based VL embedding. Extensive simulations demonstrate statistically significant gains over the evaluated baselines, with MARViC improving the VNR acceptance ratio and the SP's revenue-to-cost ratio by approximately 59% and 10%, respectively, relative to their average performance.

Department(s)

Computer Science

Publication Status

Full Text Access

Keywords and Phrases

Data center; Entropy; Network virtualization; Resource allocation; VIKOR; Virtual network embedding

International Standard Serial Number (ISSN)

1383-7621

Document Type

Article - Journal

Document Version

Citation

File Type

text

Language(s)

English

Rights

© 2026 Elsevier, All rights reserved.

Publication Date

01 Dec 2026

Link to Full Text

Share

 
COinS