An Agentic AI-Driven RAN Optimization Framework for Real-Time CCO Using 3D Beam Tilt Adaption

초록

Beam tilting is critical for Coverage capacity Optimization (CCO) in 5G/6G networks, yet uncoordinated tilt decisions across gNBs lead to unstable interference and degraded cell-edge performance. This paper presents the Multi-Agent Cooperative Tilt Optimization (MACTO) framework, which adopts an Agentic AI paradigm to enable coordinated 3D beam tilt adaptation through agent-to-agent (A2A) interaction. By jointly reasoning over vertical and horizontal tilt adjustments, MACTO improves SINR, cell-edge throughput, and interference stability, supporting AI-native RAN evolution.