AI-based NFV Management and Orchestration

Abstract

The advent of Software-Defined Networking (SDN) and Network Functions Virtualization (NFV) has revolutionized the networking industry by transforming traditional network deployment, operations, and management. SDN enables centralized management by separating control and data forwarding functions, while NFV enhances network flexibility by virtualizing network services as software components called Virtual Network Functions (VNFs). The integration of these technologies facilitates rapid, dynamic resource allocation and service deployment, thereby maximizing network flexibility and management efficiency. However, despite these advancements, the increased complexity makes effective management more difficult. This thesis focuses on developing an AI-based NFV Management and Orchestration (NFV MANO) system designed to automate and optimize the management of VNFs. The proposed system incorporates various AI-driven NFV management functions, including VNF deployment, auto-scaling, service function chaining, and consolidation. These functions are essential for effectively managing dynamically changing network environments. The primary research objective is to achieve a zero-touch network management system requiring no human intervention. To this end, the NFV-AI module within the system generates optimal policies for NFV management tasks through machine learning algorithms such as Graph Neural Networks (GNN) and Deep Reinforcement Learning (DRL). Additionally, the system includes a digital twin to replicate and simulate the NFV environment, providing a proactive learning environment for AI models. The evaluation of the proposed AI-based NFV MANO system demonstrates significant improvements in service quality, resource utilization, and operational efficiency. This research contributes to the field by presenting a comprehensive approach to integrating AI with NFV management, paving the way for fully autonomous network management systems.