Profit-Maximizing Service Function Chain Embedding in NFV-Based 5G Core Networks

Abstract

Network Function Virtualization (NFV) is a promising technology to make 5G networks more flexible and cost-efficient. With NFV, a 5G network service is implemented as several Virtual Network Functions (VNFs) that run on general machines, called a Service Function Chain (SFC). A recent survey has revealed that when multiple VNFs are colocated in the same machine, contention for shared physical resources will occur and hence degrade the throughput of a VNF and finally increase its processing delays by 50%, as compared to it runs in isolation. However, prior works fail to capture this important characteristic because they simply treat machines as a resource pool without any resource contention happening, making their approach inapplicable to the SFC embedding problem when resource contention is taken into consideration. To bridge that gap, in this paper, we study a contention-aware QoS-guaranteed SFC embedding problem and formulate it as an Integer Non-Linear Programming (INLP) under a couple of constraints. Given the formulated problem is challenging to solve due to high complexity, we propose a low-complexity approach, which can achieve a near-optimal result in a reasonable time. Numerical results reveal that the proposed approach has advantages in profit, delay, and execution time compared with other state-of-the-art approaches.