Abstract
Telecommunication network operators face growing pressures from rapidly increasing user numbers, highly diverse operating conditions, and a persistent mandate to reduce costs. In response, the industry has moved beyond initial digital transformation efforts and is now actively leveraging AI- and ML-based intelligence to drive higher levels of automation on the journey toward fully autonomous networks. While these approaches have demonstrated clear value for specific use cases, network operations remain a complex adaptive system composed of thousands of tightly coupled functions. Despite strong interest in large language model (LLM)–based approaches as a means to break down operational silos, achieving holistic, self-adaptive network behaviour remains an open challenge. The vision of truly autonomous networks (networks which are capable of reasoning, learning, and adapting across the full operational lifecycle) has yet to be fully realized. In this keynote, I will outline a potential path toward autonomous networks and present recent research results that demonstrate meaningful progress toward this goal.
