Resilient Placement of VNFs and Distributed MANO Components in a WMN-based NFV Infrastructure
Conference: Mobilkommunikation - 26. ITG-Fachtagung
05/18/2022 - 05/19/2022 at Osnabrück
Proceedings: ITG-Fb. 304: Mobilkommunikation – Technologien und Anwendungen
Pages: 6Language: englishTyp: PDF
Authors:
Frick, Gregor (Research Group for Telecommunication Networks, Frankfurt University of Applied Sciences, Frankfurt/Main, Germany Centre for Security, Communications and Network Research, University of Plymouth, Plymouth, UK)
Trick, Ulrich; Lehmann, Armin (Research Group for Telecommunication Networks, Frankfurt University of Applied Sciences, Frankfurt/Main, Germany)
Ghita, Bogdan (Centre for Security, Communications and Network Research, University of Plymouth, Plymouth, UK)
Abstract:
Wireless Multi-Radio Outdoor-Routers are building a Wireless Mesh Network (WMN), which is used as an emergency communication infrastructure in case a disaster destroyed the existing communication system. A dynamic service provisioning shall be achieved by integrating Network Function Virtualisation (NFV) into the network. A distributed management and orchestration (MANO), which operates in-band of the network, controls the NFV infrastructure (NFVI). The primary goal of the distributed MANO consists of improving the network resilience to changes affecting the NFVI, namely node and link failures. For this aspect, the position of the Virtual Network Functions (VNFs), implementing a functional service component, as well as the position of the components building the distributed MANO within the WMN-based NFVI is essential. For this purpose, this publication proposes a concept for the resilient placement of the VNFs and distributed MANO components required within the WMN-based NFVI. Furthermore, a mathematical model of the network and an Integer Linear Programming (ILP) based optimisation target for implementing the concept are proposed. The optimisation target considers the connectivity of NFVI-Nodes at the placement by preferring well-connected NFVI-Nodes and avoiding poorly-connected ones. If required due to limited hardware resources, priorities among the different kinds of network functions are considered. Multiple existing solvers are used on the optimisation target to evaluate their performance in terms of their required time for providing an optimal solution based on different network sizes.