期刊
JOURNAL OF NETWORK AND COMPUTER APPLICATIONS
卷 197, 期 -, 页码 -出版社
ACADEMIC PRESS LTD- ELSEVIER SCIENCE LTD
DOI: 10.1016/j.jnca.2021.103269
关键词
5G; User Plane Function; Service Function Chaining; ILP; Simulated annealing
类别
资金
- Agencia Estatal de Investi-gacion of Ministerio de Ciencia e Innovacion of Spain [PID2019-108713RB-C51/AEI/10.13039/501100011033]
This study addresses the placement and chaining problem of 5G User Plane Functions in a Multi-access Edge Computing environment by formulating a multi-objective ILP model and introducing two heuristic solutions for enhancing solution efficiency.
This study attempts to solve the placement and chaining problem of 5G User Plane Functions (UPFs) in a Multi-access Edge Computing (MEC) ecosystem. The problem is formalized as a multi-objective Integer Linear Programming (ILP) model targeted at optimizing provisioning costs and quality of service. Our model takes into account several aspects of the system such as UPF-specific considerations, the Service Function Chain (SFC) requests topology (single and multiple branches), Virtual Network Function (VNF) order constraints, service demands, and physical network capacities. Since the formulated problem is NP-hard, two heuristic solutions are devised to enhance solution efficiency. Specifically, an algorithm called Priority and CautiousUPF Placement and Chaining (PC-UPC) and a simulated annealing (SA) meta-heuristic are proposed. Through extensive simulation experiments, we evaluated the performance of the proposed solutions. The results revealed that our solutions outperformed the baselines (i.e., two greedy-based heuristics and a variant of the classical SA) and that we had obtained nearly optimal solutions with significant reductions in running time. Moreover, the PC-UPC algorithm can effectively avoid SFC rejections and improve provisioning costs by considering session requirements, current network conditions, and the effects of VNF mapping decisions. Additionally, the proposed SA approach incorporates several mechanisms (e.g., variable Markov chain length and restart-stop) that allow the improvement of not only the quality of the solutions but also their computation time.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据