At the 5G-A Core Network Summit held during MWC Barcelona 2025, Eric Luo, president of MAE domain of the Cloud Core Network product line, delivered a keynote speech titled "Intelligent O&M Accelerates Core Network Transition to AN L4." He introduced the intelligent O&M agent solution and deterministic high stability solution delivered by Huawei's core network ICNMaster, which can help operators accelerate their digital and intelligent transition to AN L4.
In recent years, the autonomous network industry is growing rapidly. The advent of new technologies, such as large models and intelligent agents, propels the industry to achieve a significant leap forward from "L4 is coming" to "L4 is on." In November 2024, TM Forum and 73 industry partners jointly released the AN L4 industry blueprint. It defines the characteristics, objectives, and evolution path of AN L4. The major characteristic of AN L4 is the wide application of intelligent agents, aiming to achieve automatic closed-loop. There are two phases to AN L4. Phase 1, starting from 2025 to2027, targets the single-domain autonomy. Phase 2, spanning from 2028 to 2030, aims to realize cross-domain autonomy.
Among the20 high-value scenarios defined by TM Forum in the first phase of AN L4, fault management and network change are of great concern to the core network field. The objective is to achieve automatic closed-loop throughout the whole process covering awareness, analysis, decision, and execution. Additionally, the industry blueprint highlights high stability as a unique key feature of core networks. The high stability evaluation standards were debuted to comprehensively evaluate and score operators' network reliability from seven dimensions. More than 20 leading operators participated in the evaluation.
Huawei launched the ICNMaster solution to guarantee automatic closed-loop and high reliability for high-value scenarios of core network AN L4. This solution has reshaped the core network O&M mode and high stability capabilities, and helps global operators accelerate their digital and intelligent transition to AN L4.
Reshaping O&M: Transitioning from Expert-Tool Collaboration to Agent-based Self-O&M for Automatic Closed-Loop in High-Value Scenarios
As cloud native technologies advance and new services emerge, core networks become more complex, posing significant O&M challenges. Especially in recent years, the increase of O&M manpower cannot keep up with growing O&M workload. The traditional experts+tools mode cannot meet the O&M requirements.To address these challenges, Huawei's ICNMaster solution introduces intelligent agents to reshape the core network O&M mode, making O&M simple and reliable.
Technologies including in-depth intent understanding, long text multi-round dialogue, long short-term memory, and domain-specific expertise knowledge graphs help to realize natural language interaction.
Complex tasks can be planned and executed by using chain-of-thought (CoT) reasoning(empowered by expert experience) and model self-reflection and evaluation.
A powerful technical base built on Huawei Pangu model, DeepSeek model, and professional models enables industry-leading performance and accuracy.
ICNMaster automatically executes or assists in key activities of different O&M roles throughout operators' processes. For instance, a monitoring and troubleshooting process typically requires three roles: a monitoring engineer, a core network engineer, and a decision-maker. With the ICNMaster's O&M agents, activities traditionally carried out by monitoring and O&M engineers can now be automated, and decision-makers receive valuable assistance in making decisions. Currently, the ICNMaster's O&M agents are being utilized in the production processes of China Mobile across multiple provinces. These agents have enhanced the efficiency of handling complaint and alarm tickets by over 60%, and improved analysis accuracy to over 90%.
Reshaping High Stability Capabilities: From Best-Effort to Deterministic, Building a High Stability Foundation for Core Networks
Mr. Luo explained that the advancement of high stability capabilities can be classified into five phases. Phase 1.0 ensures that the system does not break down when a network element (NE) becomes faulty. Phase 2.0 implements network fault survival. Phase 3.0 guarantees uninterrupted service running. Phase 4.0 focuses on deterministic service recovery. And Phase 5.0 achieves always-on services. Currently, Huawei's high stability 3.0 capabilities ensure the reliable operations of global operators' networks. However, 3.0 is just the beginning. We must strive to advance to high stability 4.0 to achieve deterministic service recovery capabilities.
Core network fault scenarios can be categorized into three layers: inner, middle, and outer layers, based on fault locations. Huawei's ICNMaster provides deterministic high stability capabilities across all these scenarios. For VNF/NFVI faults at the inner layer, ICNMaster enables automatic detection and rectification; for datacom network faults at the middle layer and peripheral device faults at the outer layer, ICNMaster provides fast recovery without requiring demarcation if there are alternative paths. If there are no paths, ICNMaster can restore services within 15 minutes.
Traditionally, exceptions are manually identified through monitored alarms and KPIs, and then demarcated and rectified. This time-consuming process leads to KPI deterioration, severely affecting services. Huawei's demarcation-free fast recovery solution revolutionizes this process by enabling automatic exception detection, service recovery, and fault demarcation.
This solution monitors KPIs in real time from multiple dimensions. When an exception is detected, for example, KPI deterioration exceeding 5%, the solution automatically initiates path selection and service restoration before a fault is reported.
During troubleshooting, the solution detects path health in real time using an AI-powered bidirectional path health detection algorithm and automatically suggests the optimal path based on a multi-objective optimization algorithm. This helps O&M personnel rectify faults in one-click mode.
Following recovery, the cross-layer fault diagnosis agent automatically demarcates the fault within 10 minutes.
AN L4 is on. With advanced agent technology and the concept of deterministic high stability, Huawei's ICNMaster is reshaping the core network O&M mode and high stability capabilities, empowering global operators to achieve efficient O&M and high network stability, accelerating the evolution to AN L4.