#USA #Huawei #AI Network #Peking #Next-Gen Core

Huawei's Vision for an AI-Driven Next-Generation Core Network Revolution

As communication networks evolve, mobile networks have traditionally paved the way for early interaction through voice calls, text messaging, and internet access. However, the emerging era of AI agents necessitates a significant transformation, where networks must handle an ever-increasing number of AI-driven interactions. This evolution demands that the core network architecture understand intentions, make intelligent decisions, and facilitate complex task completion driven by AI.

Huawei believes that a native AI core network is essential for this transformation. By embedding AI into interfaces, features, and processes, operators can shift from merely providing connectivity to delivering intelligent network functions tailored for the next generation of digital services. According to Huawei's report, "Intelligent World 2035," the global count of AI agents may reach a staggering 900 billion by the year 2035—approximately 100 times the world’s human population. To accommodate this surge, core network architecture must be equipped to support the long-term growth of new connections and services.

With the expansion of AI terminals and applications, agent-based services are predicted to become common scenarios within mobile networks. This will drive a shift from pure data transmission to enabling task-oriented tokens. Huawei has identified three typical service scenarios requiring the reliable operation of AI-driven systems:

1. Agent Over New Radio (AoNR) for task latency assurance: As AI devices face limitations regarding computational power, battery capacity, size, and cost, managing complex latency becomes imperative. AoNR is designed to consistently deliver an excellent user experience while keeping task latency from start to finish beneath demanding thresholds.

2. Wide-Area Computing Network (WCN) for enhanced computational efficiency: WCN can connect computing resources via mobile networks, facilitating computation scheduling based on tasks and standardized computing services.

3. Agent Computing Network (ACN) for collaboration assurance: Most complex tasks will involve several AI agents collaborating across multiple domains, enterprises, and ecosystems, rather than a single agent limited by modeling capabilities or contextual constraints. ACN will enable efficient inter-agent cooperation across domains through trusted digital identities, skill-based agent discovery and routing, and dynamic task-led network creation.

To bridge the gap between terminals and networks, next-generation core network architecture must incorporate three key characteristics:

• - Interfaces (from signaling to intention): Intent-based interfaces will allow terminals to communicate tailored task requirements—including network, computing, and experience needs—directly to the network.
• - Functions (from C/U planes to C/U+DAP planes): The introduction of a dedicated Data and AI Plane (DAP) alongside traditional control and user planes will act as an intelligence layer, interpreting task intentions and dynamically aligning resources in real time.
• - Procedures (from AI overlay to native AI): AI decision-making will be inherently integrated into fundamental network processes for seamless coordination of connections and computations.

Huawei outlines a two-phase development plan for AI-driven core networks. The first phase focuses on enhancing customer experiences by layering AI functionalities. The second phase envisions a next-gen core network enriched with AI capabilities, which facilitates what Huawei terms "enhanced connectivity," where task-oriented AI services drive task completion, along with "beyond connectivity" services providing functionalities that extend beyond communication, such as computing services and enabling agent collaboration.

For the network to seamlessly adapt to these new services, a comprehensive restructuring of network architecture will be required, especially concerning interaction mechanisms, data architecture, and computational architecture. Moreover, transitioning to native AI networks necessitates extensive collaboration across the entire ICT ecosystem, including terminals, chips, and networks. Through synergy mechanisms between terminals and networks and the advanced architecture of next-gen core networks, new connections and services can emerge, unlocking fresh business opportunities in the intelligence era.