Virtual Network Function Placement and Routing: Formulations and Solutions

Artikel i vetenskaplig tidskrift, 2026

The Virtual Network Function Placement and Routing Problem (VNFP-RP) represents a fundamental challenge for achieving efficiency and scalability in Beyond Fifth-Generation (B5G) and Sixth-Generation (6G) networks. It encompasses both the placement of Virtual Network Functions (VNFs) on available computational resources and the routing of traffic through them. These routes must follow the order defined by Service Function Chains (SFCs) within Network Function Virtualization (NFV)-enabled environments. In this work, we present a complete and reproducible implementation of an established Integer Linear Programming (ILP) model adapted from the literature, integrated into a flexible and publicly available experimental testbed for the research community. This testbed includes multiple network topologies, workload generators, and execution scripts, enabling fair benchmarking, systematic evaluations, and future extensions. Using four real-world topologies from the Survivable Network Design Library (SNDLib) under varying workload levels, we optimize the VNFP-RP under four distinct objective functions: node energy consumption, number of active nodes, number of allocated VNFs, and aggregate system latency. The results reveal substantial differences in computational behavior, with execution times ranging from under 0.1 seconds to more than 6 hours, driven by differences in objective functions and network topologies. These findings highlight how modeling choices impact both runtime and placement structure, providing valuable insights into the trade-offs among efficiency, responsiveness, and resource utilization. Ultimately, this study emphasizes the importance of flexible and adaptive strategies to guide decision-making in the design and operation of B5G and 6G networks.