Design and Development Of A 5G NTN Platform For Optimal AI-Assisted VNF Allocation

Abstract

Network Function Virtualization and Software Defined Networking are considered key enablers of the fifth-generation mobile networks. As research extends beyond 5G, the integration of mobile and Non-Terrestrial Networks (NTN) has gained importance to improve the flexibility, agility and cost-effectiveness of network service delivery and operations. One significant challenge in integrating 5G and NTN is the development of an emulation framework for the designing and implementation of management and orchestration algorithms. This is particularly relevant for optimising Virtual Network Function (VNF) allocation in such distributed environments. A comprehensive 5G NTN emulation environment was designed and developed for this purpose in this thesis, utilizing open-source software including Open Air Interface for 5G System and Kubernetes for orchestration. This setup enables the implementation of automatic orchestration algorithms for dynamic VNF allocation. A comprehensive examination and validation of the environment’s characteristics is presented. A key contribution of this thesis is the formulation of the VNF Allocation Optimization Problem, along with a thorough analysis of the variables required. The thesis also provides an assessment of the possible evaluation services and the implementation approaches considering the architectural paradigm and decision timeframe. As the culmination of this research, a proof-ofconcept illustrates the necessity and feasibility of using Artificial Intelligence-based algorithms to understand and solve the optimisation problem in such a complex environment.