Network Architecture for Machine Learning: A Network Operator's Perspective

Data-driven network design suggests that substantial technology advances of 5G and 6G networks will be enabled with enhanced automation, intelligence, and user-experience-focused capabilities. Network operators need to upgrade the standard network models by applying machine learning (ML) to address the complexities of next-generation network deployments. This article explores the role of ML and its interplay with wireless communications networks to develop the next-generation network architecture. A use case scenario for self-configuration of radio-access-network-based notification areas (RNAs) for effective resource management is analyzed to exemplify the proposed architecture where a paging load reduction of 64 percent is observed in the resulting RNA clusters. A conceptual framework for RNA configuration and management enabling a broader perspective toward an ML-driven hybrid self-organizing network is discussed to improve the signaling load to attain reduced latency and improved network capacity.