Beyond Automation to Autonomy: How AI Agents Are Reshaping 6G Network Management

Existing network management has relied on automation based on the FCAPS framework. In the 5G era, the standardization of MDA(Management Data Analytics) and O-RAN's RIC(RAN Intelligent Controller) laid the groundwork for intelligent operations, but there remains a limitation in that they still depend on predefined procedures and static integration logic.

Whenever the network environment changes or new vendor equipment is added, manual reconfiguration or custom development is required, and this makes E2E(End-to-End) automation difficult due to fragmented interfaces across domains, generations (4G/5G/6G), and vendors. As the evolution to 6G progresses, services become more diverse and network density increases, making autonomous decision-making beyond simple automation essential.

Agentic Network Management refers to the ability to autonomously carry out operational goals while minimizing human intervention. Unlike existing approaches that follow fixed workflows, this interprets high-level intent, establishes multi-step plans, and adjusts actions according to feedback.

The key characteristics of agentic operations are as follows:

• Intent Interpretation: Translating abstract operational goals into concrete management actions
• Capability Discovery: Dynamically exploring available management services and functions without being fixed to a specific interface
• Reasoning and Planning: Decomposing complex tasks, coordinating actions across domains, and planning multi-step procedures
• Adaptive Execution: Reasoning about outcomes through feedback and executing alternatives according to the situation