Joint VNF Placement and CPU Allocation in 5G

TLDR

Network slicing enables 5G to support diverse services flexibly and swiftly. The study aims to jointly optimize VNF placement across physical hosts and CPU allocation among co‑located VNFs. A queuing‑based system model is developed and a fast, efficient solution strategy is proposed to generate near‑optimal placement and CPU allocation decisions. Evaluation across realistic scenarios shows the approach consistently outperforms state‑of‑the‑art methods while approaching optimal performance.

Abstract

Thanks to network slicing, 5G networks will support a variety of services in a flexible and swift manner. In this context, we seek to make high-quality, joint optimal decisions concerning the placement of VNFs across the physical hosts for realizing the services, and the allocation of CPU resources in VNFs sharing a host. To this end, we present a queuing-based system model, accounting for all the entities involved in 5G networks. Then, we propose a fast and efficient solution strategy yielding near-optimal decisions. We evaluate our approach in multiple scenarios that well represent real-world services, and find it to consistently outperform state-of-the-art alternatives and closely match the optimum.