The cellular communication business stands on the threshold of profound transformation, pushed by the fast evolution of synthetic intelligence (AI). The widespread adoption of open-source AI massive fashions, equivalent to DeepSeek, is making highly effective capabilities accessible to extra folks, whereas the rise of AI Agent companies, exemplified by frameworks like Manus, is shifting AI’s function from passive “advisor” to energetic “executor.” On this quickly rising Cellular AI period, ZTE positions itself on the vanguard, pioneering the deep integration of AI with radio entry networks (RAN) by means of its AIR RAN (AI Reshaped RAN) answer. At MWC Shanghai 2025, Tang Xue, Vice President of ZTE, shared this imaginative and prescient throughout a keynote on the GSMA’s 5G Enterprise Summit.
“The deep convergence of cellular networks and AI is ushering in new service paradigms,” Tang emphasised. “AIR RAN leverages heterogeneous computing, with the AIREngine—an clever computing board embedded throughout the BBU—because the enabler of community native-AI. By integrating complete digital upgrades throughout antennas, transmission and energy techniques, it transforms legacy RAN into an infrastructure the place AI and RAN are essentially intertwined.”
AIR RAN harnesses the community’s native-AI capabilities—together with clever service identification, knowledge analytics and deep studying, in addition to service-assured useful resource orchestration—enabling operators to reinforce person expertise whereas bettering community effectivity.
To reinforce person expertise, AIR RAN implements a exact closed-loop assurance system. This technique achieves its targets by means of focused subscriber acquisition, correct notion and analysis of person expertise, and ensures that personalised wants, significantly for premium customers in high-value situations, are constantly met. By integrating key situations, service add-on packages and person classification, operators can considerably enhance high-value buyer satisfaction and drive substantial income progress by means of focused service enhancements.
To enhance community effectivity, AIR RAN leverages exact service notion to deploy refined methods. These embody superior energy-saving optimization, load-based dynamic beam adaptation and service-aware grievance root trigger evaluation. This holistic method not solely elevates person notion but additionally considerably boosts spectrum, power and operation effectivity. The outcomes embody elevated cell capability, diminished community power consumption, streamlined O&M processes, and substantial Opex financial savings.
“Constructing on AIR RAN, we introduce AgentGuard – the progressive answer for human-machine interplay revolution within the cellular AI period,” said Tang. “We’re witnessing a pivotal transition from cellular web to cellular AI. On this new period, conventional app interfaces are evolving into AI Brokers, showing as digital assistants and embodied brokers like robots. AgentGuard empowers the deep integration of 5G-A and AI, offering enhanced assurance for brand new AI companies.”
At MWC Shanghai 2025, ZTE showcased a dwell demonstration of how AgentGuard empowered AI Agent to supply companies equivalent to ordering espresso, producing opinions and posting to social media with single voice command. The high-capacity 5G-A community assured that the AI Agent carried out flawlessly amidst dense crowds.
With out manually opening any app, a single natural-language command triggered the AI Agent to seamlessly deal with geolocation, intent parsing, service provider matching and cost.
Tang emphasised the implications, “2025 would be the ‘12 months of the AI Agent Explosion.’ Whereas offering service like ‘espresso at your command,’ these light-weight interactions impose unprecedented calls for on cellular networks. To stability gadget energy, price, and effectivity, AI Brokers depend on end-cloud collaboration: units deal with preliminary intent understanding, whereas cloud-based LLMs execute multi-step duties. This entails the fixed trade of multimodal knowledge (voice/picture/video) and drives interactions demanding ultra-high uplink bandwidth and ultra-low latency.”
“Analysis confirms that guaranteeing clean multi-step AI Agent duties requires networks to ship ubiquitous 20Mbps uplink capability,” Tang said. “AgentGuard is devoted to making sure flawless AI Agent experiences.”
Conventional static assurance insurance policies should not well-suited for AI Brokers. “When a digital assistant performs duties like ordering espresso—choosing distributors, evaluating costs, inserting orders, processing funds—every step dynamically modifications its community wants.”
AgentGuard pioneers “one-step-one-policy” precision assurance with three core improvements:
-Clever Service Recognition: Base stations use neural networks to deeply correlate and acknowledge sequences of actions carried out by digital assistants—from cloud planning to native execution—figuring out particular agent sorts and targets (e.g., ordering meals, reserving tickets).
-Composite Assurance Insurance policies: Adapting to generative activity dynamics, the answer dynamically generates optimum community insurance policies and allocates exact assets for every activity step to make sure dependable, clean execution.
-4-Tier Collaborative Assurance: A closed-loop mechanism spanning cell-level, base station cluster-level, network-level, and device-network-level collaboration delivers deterministic community experiences.
“We’re dedicated to delivering ubiquitous clever experiences,” Tang affirmed. “Everytime you name upon your sensible assistant, relaxation assured that 5G-A and AgentGuard work silently behind the scenes, eliminating worries about community stability or complexity.”
Additional elaborating on AIR RAN’s differentiation, Tang famous, “The deep integration of RAN and AI may be understood from two views: ‘AI for RAN’ and ‘RAN for AI’.” Whereas many present options focus totally on “AI for RAN” to drive upgrades and improve community worth, AIR RAN additionally pioneers “RAN for AI,” enabling RAN to help edge AI deployments and increasing its strategic function.
Trying forward, ZTE plans for AgentGuard to evolve in tandem with multimodal AI brokers involving voice, video and picture inputs. The answer dynamically allocates assets with latency ensures tailor-made to the interplay kind.
“5G-Superior serves as important infrastructure for the Cellular AI period,” Tang stated. “ZTE drives 5G-Superior community evolution by means of twin pillars: strengthening community foundations and enabling deep AI integration. This transformation empowers the cellular AI period by means of clever infrastructure evolution.”
