Intent-based Networking for QoS-aware Cloud and Transport Network Management based on Graph Neural Networks

The complexity and various goals of beyond 5G / 6G networks, characterized by heterogeneous and disaggregated Radio Access Network (RAN) components, controllers and Service Management and Orchestration (SMO) architectures, by the introduction of additional access network technologies, by advanced sensing and positioning technologies and by ever-more-specialized, privately-operated campus networks necessitate advanced network management approaches. To meet these demands, it is essential to enhance network management automation and to simplify network management interfaces. These two aspects serve as the foundation for the Intent-based Networking (IBN) - based Network and Edge Data Management Interface (NEMI) Network Management System (NMS) encapsulating dedicated AI/ML models into Intent Management Functions (IMFs) for enabling advanced network management use cases such as energy-efficiency-, Quality of Service (QoS)-and ressource-optimizations. By integrating advanced Artificial Intelligence (AI)/Machine Learning (ML) capabilities at various levels of the network, including RAN, Core, transport network, and Cloud new levels of network management automation are realized. This work provides an insight into the Fraunhofer FOKUS toolkit Network and Edge Data Management Interface (NEMI)'s Intent-based network management system's implementation and its integration of Graph Neural Networks (GNN) and evaluates its performance for autonomous transport network and cloud infrastructure management and optimization.