Efficient Resource Allocation for NextG O-RAN VNFs Based on Systematic Characterization
Haoxin Sun, Mattia Lecci, Javier Rivas, Germán Corrales Madueño, Francisco J. Garcia, Francisco Muro, Francisco Crespo, Sergio Fortes, Raquel Barco · 2025
The deployment of Virtualized Network Functions (VNFs) in Open Radio Access Network (O-RAN) architectures for beyond 5G (b5G) and 6G networks presents a fundamental challenge in resource dimensioning. Network operators must accurately provision computational, memory and network resources to ensure service reliability while maximizing infrastructure efficiency at resource-constrained edge sites. Underprovisioning results in service degradation and outages, while over-provisioning reduces deployment density and increases operational costs. Despite its critical importance, the relationship between the consumption of VNF resources and operational parameters remains insufficiently characterized, dealing with uncertainties in resource planning. In this paper, we introduce SANGRE (Systematic chAracterization of NextG O-RAN VNF rEsource utilization), a framework for comprehensive evaluation of VNF resource requirements under diverse operational conditions. SANGRE employs a wraparound testing methodology to ensure measurement accuracy and repeatability in isolated environments. We implement this framework on a production-grade testbed using open-source gNodeB solutions (SRS) and perform extensive evaluations across multiple network configurations. Our analysis reveals the complex dependencies between the utilization of VNF resources and key parameters such as cell bandwidth, MIMO configuration, user density, and traffic patterns. These findings provide essential information for operators to optimize resource allocation strategies in NextG deployments.