• Smaller Small Medium Big Bigger
  • Default Helvetica Segoe Georgia Times

Yang Chaobin, Board Member, President of ICT Products & Solutions at Huawei, provided an insightful overview of artificial intelligence (AI) for network and productivity during the first day of the Mobile World Congress (MWC) in Shanghai.

Chaobin highlighted the historical progression of mobile communications, tracing its development from the rapid evolution of mobile internet and mobile video to the current era of mobile AI. The continuous integration of evolving AI technologies has significantly improved network automation. Transitioning from manual configuration to tool-assisted task automation has greatly enhanced network operations and maintenance (O&M) efficiency and the overall service experience.

As the telecom industry marks the first year of 5G-A, with generative AI (GenAI) making significant breakthroughs, AI is poised to drive networks towards an era of high-level autonomy.

Transforming Traffic and Connection Density

The rapid growth of AI-generated content (such as video) is expected to trigger a new boom in network traffic. It is predicted that traffic growth will reach a turning point in 2025, with network traffic increasing 20 times by 2030. Additionally, various AI terminals, including digital humans, smart vehicles, and industry-specific smart devices, are emerging. By 2030, it is estimated that these terminals will drive a more than tenfold increase in the number of connections, while the penetration rate of AI terminals is expected to surpass 70%, and the virtual human market is projected to reach over 270 billion RMB.

By 2030, AI is expected to be applied to tens of thousands of industry segments, with the number of AI assistant users reaching 1 billion and users of digital livestreaming, smart home, and embodied AI reaching 100 million. These advancements place new demands on business experience. The first encompasses differentiated experience operation, which includes end-to-end, ultra-low latency and deterministic assurance. The second encompasses more advanced network automation capabilities, including user self-service features, natural semantic interaction between humans and machines, and self-closure capabilities for network service scenarios.

Telecom Foundation Models

Over the past two decades, network operators have seen their capital expenditures (CapEx) decrease in the form of shared revenue, while operational expenditures (OpEx) have remained high. Major components of OpEx include installation and maintenance, electricity costs, and network optimization, accounting for 58% of total expenses.

Introducing telecom foundation models has evidently reduced OpEx by addressing challenges like network complexity, precision, and efficiency. These models can reshape the O&M paradigm through applications such as copilots and agents, providing new modes of operation and maintenance.

The high-level autonomous network framework enabled by telecom foundation models features a full-stack architecture. This architecture includes a:

  • Business Operation Layer: Facilitates agile service provisioning and adjustment, serving as an intelligent interface for users to access carrier services and realize business value.
  • Network O&M Layer: Functions as the intelligent core for operational management, conducting closed-loop operations and training foundation models to generate optimized policies for network maintenance.
  • Network Resource Layer: Manages resources through closed-loop management and real-time execution, akin to an intelligent trunk with capabilities for executing network policies efficiently.

Huawei’s Telecom Foundation Model Boosts Network Productivity

At the recent MWC Barcelona 2024, Huawei launched the Telecom Foundation Model, enhancing its ADN solution capabilities. This advancement enables efficient closed-loop network maintenance through intelligent analysis and inference, ensuring dynamic and balanced service experiences with analysis, and fast multi-objective optimization. Consequently, service rollout and monetization are accelerated through intention exchange and automatic API generation.Focusing on high-value scenarios in network maintenance, experience assurance, and service enablement, Huawei has developed five role-oriented copilots and five scenario-oriented agents. These copilots feature natural semantic interaction, intelligent knowledge Q&A, and O&M assistance, reducing manual tasks and improving operational efficiency across various roles.

The five scenario-oriented agents address complex O&M scenarios by breaking down tasks, invoking appropriate tools and APIs, and autonomously resolving issues. This approach enhances service quality, reduces O&M costs, and improves overall efficiency.

Three Application Scenarios


Addressing fault management, Huawei’s Telecom Foundation Model encompasses FaultSpirit, a comprehensive solution tackling various aspects of network fault handling.

During the fault identification phase, the model collaborates with a high-precision digital twin, utilizing multi-layer and multi-dimensional network. This synergy ensures early detection and accurate localization of faults within the network infrastructure.

Addressing the fault diagnosis phase, the Telecom Foundation Model employs sophisticated chain-of-thought intelligent analysis capabilities. This approach swiftly pinpoints the root causes of issues, significantly reducing the number of alarms.

To combat software commissioning faults, which account for 40% of identified issues, FaultSpirit automatically triggers the appropriate reactive tools or APIs to rectify these faults, streamlining the troubleshooting process.

Handling the remaining 60% of hardware faults requires the expertise of copilots. These AI-assisted systems provide site engineers with precise fault details and actionable rectification suggestions, facilitating quicker resolutions. As a result, the duration for processing trouble tickets is slashed sharply, enhancing operational efficiency and minimizing network downtime.


In the domain of wireless network optimization, OptimSpirit comprehensively evaluates objectives such as energy efficiency, performance enhancement, and user experience. Factors like data rate, interference, coverage, and power consumption are meticulously analyzed and optimized.

Powered by the Telecom Foundation Model and an integrated RAN digital twin system, OptimSpirit autonomously formulates and implements policies aimed at maximizing energy savings without compromising user satisfaction. This approach boosts network efficiency by observably reducing power consumption and enhancing cell-edge data rates, thus, maximizing overall network capabilities and user experience.


Addressing enterprise campus service-enabling scenarios, ProvSpirit plays a crucial role in streamlining service provisioning and operational management.

During the provisioning of enterprise campus private networks, ProvSpirit automatically generates tailored SLA configurations. This eliminates the need for on-site visits, reducing service provisioning time.

For complex campus networks, ProvSpirit facilitates collaborative O&M through integrated operation screens, O&M interfaces, and remote device management capabilities. This transformation enhances operational efficiency by enabling single-person management across multiple domains, significantly reducing the mean time to repair (MTTR).

Moreover, ProvSpirit excels in remote fault rectification, addressing most of issues without requiring physical visits to campus locations. This capability enhances enterprise O&M capabilities, fostering a more agile and responsive service environment.

These applications highlight Huawei's innovative approach to leveraging AI and advanced network models to optimize operations, enhance user experience, and drive efficiency across diverse network environments.

Chaobin concluded his speech by encouraging stakeholders to, "Seize the new opportunities in the mobile AI era, use intelligence to empower networks, implement innovative practices to accelerate the development of high-level autonomous networks, and make networks more productive."

By Clarissa Garcia, Journalist of Telecom Review Group

Pin It