DeViNE: A Decentralized Virtual Network Embedding Algorithm
Farzad Habibi, Juncheng Fang
TL;DR
DeViNE tackles the scalability and robustness limitations of centralized VNE by distributing embedding tasks across multiple randomly chosen leaders and using a ring-based election to select the best embedding based on a revenue–cost metric. Each leader performs a BFS-based local embedding on its portion of the physical network, and a lightweight message-passing protocol consolidates candidates without a central controller. The approach is shown to improve acceptance rates and the revenue-to-cost ratio relative to centralized baselines, demonstrating the practical viability of fully decentralized VNE for large-scale or edge-oriented deployments. This decentralized framework enhances robustness against node failures and DoS threats while preserving embedding efficiency, making it well-suited for distributed data centers and edge networks.
Abstract
Virtual Network Embedding (VNE) is a technique for mapping virtual networks onto a physical network infrastructure, enabling multiple virtual networks to coexist on a shared physical network. Previous works focused on implementing centralized VNE algorithms, which suffer from lack of scalability and robustness. This project aims to implement a decentralized virtual network embedding algorithm that addresses the challenges of network virtualization, such as scalability, single point of failure, and DoS attacks. The proposed approach involves selecting L leaders from the physical nodes and embedding a virtual network request (VNR) in the local network of each leader using a simple algorithm like BFS. The algorithm then uses a leader-election mechanism for determining the node with the lowest cost and highest revenue and propagates the embedding to other leaders. By utilizing decentralization, we improve the scalability and robustness of the solution. Additionally, we evaluate the effectiveness of our fully decentralized algorithm by comparing it with existing approaches. Our algorithm performs $12\%$ better in terms of acceptance rate and improves the revenue-to-cost ratio by roughly $21\%$ to compared approaches.