Optimal Service Provisioning Schemes for Multi-Layer Fog Networks
Anton Riedl, Nazli Siasi · 2021 IEEE Globecom Workshops (GC Wkshps) · 2021
The deployment of network function virtualization (NFV) in fog networks allows network providers to offer dynamic, delay-sensitive services on top of general purpose hardware. A key challenge is the provisioning of service function chaining (SFC), i.e., the placement of requested virtual network functions (VNF) across the nodes, in ways that make efficient use of the sometimes limited resources while meeting stringent quality of service (QoS) requirements.In this paper we propose a novel optimization scheme for placing VNFs in the network and we apply it to a multi-layer fog architecture composed of heterogeneous global and local fog nodes. The algorithm is based on integer linear programming (ILP) and gives particular consideration to multi-mapping and instance-sharing. Results show that SFC with VNF multi-mapping and instance-sharing generally yields the best resource utilization, along with achieving short delays and high admission rates.