#USA #Huawei #AI Network #Peking #Core Network

Transforming Mobile Networks with AI Integration

Mobile networks have traditionally been utilized for voice calls, text messaging, and internet access. However, as we enter an age dominated by autonomous AI agents, there is an urgent need for a new kind of network infrastructure capable of managing vast numbers of these agents. This evolution looks to redefine the core architecture of mobile networks to not only connect users but to understand their intentions, make intelligent decisions, and ensure the successful execution of AI-directed tasks throughout their life cycle.

Huawei has identified that the key to this transformation lies in a core network that inherently supports artificial intelligence (AI). By integrating AI directly into interfaces, functions, and procedures, network operators can transition from simply providing connectivity to delivering intelligent networking capabilities for next-generation digital services. According to Huawei's report titled "The Intelligent World 2035," the number of AI agents globally could reach 900 billion by 2035—around 100 times the world's population. As the demand for these agents continues to increase, the core network architecture must be prepared to support sustained growth in new connections and services.

The growing array of AI-enabled terminals and applications is set to make agent services typical scenarios in mobile networks. This represents a shift from merely transmitting bits of data to enabling task-oriented tokens. Huawei has pinpointed three principal scenarios requiring reliable operation of services utilizing AI agents:

1. Agent Over New Radio (AoNR): This scenario focuses on ensuring low latency in task completion. Portable AI devices often face constraints related to computational power, battery life, size, and cost, making end-to-end latency a critical factor. AoNR aims to consistently deliver an excellent user experience while keeping task latency within stringent thresholds.

2. Wide-Area Computing Network (WCN): This framework connects computational resources via mobile networks, allowing for task-based computation planning and standardized computational services.

3. Agent Computing Network (ACN): Given that many complex tasks will involve multiple AI agents collaborating across various domains and ecosystems rather than relying on a single agent, the ACN ensures secure and dynamic collaboration. This architecture supports effective agent collaboration through three features: cross-domain trusted digital identity, skill-based agent search and routing, and task-driven dynamic network connections.

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

• - Intent-based Interface: This interface enables endpoint devices to communicate specific task completion requests directly to the network, tailored to individual needs, including connectivity, computational power, and user experience quality.

• - Functionality Enhancements: In addition to traditional control and user planes, the architecture introduces a separate Data and AI Plane (DAP), which serves as an intelligent layer that interprets intentions related to each task and dynamically adjusts resource availability in real-time according to current needs.

• - Procedural Integration: AI-driven decision-making processes are intricately woven into fundamental networking procedures, allowing for seamless coordination of network connectivity and computational resources.

Looking forward, Huawei envisions two phases in the evolution of the AI-controlled core network. The first phase focuses on how the network can enhance customer experiences through overlay AI functions. The second phase introduces a next-generation core network with native AI capabilities, fostering "enhanced connectivity," where task-oriented AI services ensure task completion, and "beyond connectivity," where the network provides functionalities beyond communication, such as facilitating computation services and agent collaboration.

In order for network architecture to adjust to these new services, significant restructuring is required, particularly in relation to interaction mechanisms, data architecture, and computational architecture. Furthermore, the transition to AI-powered networks demands extensive collaboration across the entire ICT ecosystem, encompassing terminals, chips, and networks. With a synergy mechanism between terminals and the network, and an advanced core network architecture, next-generation core networks can create new connections and services, unlocking new business opportunities in the intelligence era.