부산대학교 도서관

The complexity of orchestrating Beyond 5G services, such as vehicular, demands novel approaches to remove limitations of existing techniques, as these might cause a large delay in orchestration operations, and thus, negatively impact the service performance. For instance, the human-in-the-loop approach is slow and prone to errors, and closed loop control using rule-based algorithms is difficult to design, as an abundant number of parameters need to be configured. Applying Artificial Intelligence (Al)/Machine Learning (ML), in combination with Network Function Virtualization (NFV) and Software Defined Networking (SDN), seems a promising solution for enabling automation and intelligence that will optimize orchestration operations. In this article, we study the challenges in current ETSI NFV orchestration solutions for B5G C-V2X edge services; propose an Al/ML-based closed-loop orchestration framework; propose how and which Al/ML techniques can alleviate the identified challenges and what are the implications resulting from applying certain Al/ML techniques; and discuss A//ML-based system enablers for B5G C-V2X services.