Dynamic and Mobility-Aware VNF Placement in 5G-Edge Computing Environments

Abstract

The combination of both Network Function Virtualization (NFV) and Edge Computing facilitates the Quality of Service (QoS) fulfillment of latency-stringent applications by taking advantage of the proximity of the devices at the edge and the user, and provides the necessary flexibility to enable agile, cost-effective, and on-demand service delivery models for 5G environments. However, to achieve those benefits, VNFs must be deployed so that system resource utilization is optimized and service QoS is maintained. In this context, the selection of the most suitable edge nodes to deploy VNFs is a challenging research issue because multiple — and often conflicting — objectives must be considered by the VNF placement algorithm. The challenges become greater when considering the high dynamism of the edge environment. The availability of edge nodes and users’ locations vary over time, making original allocation decisions ineffective. In this paper, we propose DSP, a Dynamic Smart VNF Placement algorithm that considers the dynamic nature of a 5G edge environment. Our proposal focuses on optimizing resource utilization with an online placement approach to handle continuously arriving service requests. Furthermore, we also account for user mobility providing an approach that reduces the performance impact caused by users’ movement. Additionally, a two-level resource-sharing mechanism is provided, minimizing the cost for the infrastructure provider with low impact on QoS.