Intelligent radio access network for 5G-Advanced and 6G: key standard technologies and future evolution

5G/5G-Advanced is expected to continuously enhance key performance indicators, requiring further breakthroughs in aspects such as latency, reliability, connection density, and user experience. The limitations of traditional human-computer interactive network management, which relies primarily on manual operations, are becoming increasingly evident in terms of efficiency, accuracy, and cost. Compared to traditional optimization methods, the predictive and forward-looking capabilities of AI/ML enable the network to shift from passive response to active perception and self-optimization, achieving a transition from “monitoring-reaction” to “prediction-orchestration”. Based on the key technologies and standardization paths for radio access network (RAN) intelligence defined by 3GPP, the AI/ML model management, data collection, and interaction mechanisms were analyzed in conjunction with typical use case scenarios. For intelligent RAN in 6G, a novel architectural concept termed the “intent-driven collaborative task”was proposed. Its key implementation relied on RAN's awareness of application-layer information, task-level quality of service (QoS) monitoring, dynamic grouping and resource management, and other technologies to achieve seamless interaction between human, machine, and the carbon-silicon ecosystem in 6G networks.