etisalat UAE and Khalifa University publish AI-native 6G blueprint
etisalat UAE and Khalifa University of Science and Technology have jointly published a peer-reviewed architectural blueprint for AI-native 6G networks, setting out how intelligence can be embedded directly into the foundations of future mobile infrastructure rather than layered on top after the fact. The paper appears in the IEEE Open Journal of the Communications Society and is the third in a three-part research series the partners have completed during 2026.
The central innovation in the study is a dedicated "AI-plane" that sits alongside the user, control and management planes already defined in existing network architectures. The AI-plane gives operators, vendors and standards bodies a single reference model for how a 6G network would learn from live conditions, make decisions autonomously and act on those decisions without waiting for human intervention. The architecture distributes AI agents across four tiers, from connected edge devices up to the cloud, and specifies full lifecycle management for the AI models running within each tier.
From managed to self-managing networks
The study uses the TM Forum's six-level autonomy scale to map a progression from today's largely assisted networks towards systems capable of planning, delivering, assuring and restoring services without operator intervention. The proposed architecture targets autonomy levels three to four in validated domains, with levels four to five as the long-term objective for routine operations. Human operators are recast in an intent-setting and policy-management role rather than a hands-on operational one.
A key structural component is the network digital twin: a continuously synchronised replica of the live network used as a safe staging environment. New AI models can be trained and tested in shadow mode against production traffic before being granted authority to act on the live network, reducing the risk of autonomous decisions causing service disruption.
Marwan Bin Shakar, Chief Technology Officer at etisalat UAE and a co-author of the paper, said: "Every generation of mobile technology has made networks faster. 6G is the first that can make them think. When intelligence is part of the architecture, the network can anticipate demand, predict faults and correct itself before customers notice."
Standards alignment and commercial roadmap
The paper maps the AI-native design against the full 6G standardisation landscape, covering ITU IMT-2030, 3GPP Releases 18 to 21, the O-RAN Alliance, the AI-RAN Alliance and ETSI. A three-phase roadmap runs from research and prototyping (2026 to 2027), through standardisation and pre-commercial trials (2027 to 2029), to commercial deployment and evolution from 2029 to 2032 and beyond. The work sits within the UAE's TDRA-coordinated 6G framework, positioning etisalat and Khalifa University as contributors to global standards discussions, not merely observers.
The two earlier papers in the series covered integrated sensing and communication (July 2026) and centimetre-wave spectrum engineering for capacity, produced in collaboration with NYU Abu Dhabi (October 2026).
The broader 6G standardisation race is intensifying, with Chinese operators and vendors (China Mobile, Huawei), European incumbents (Ericsson, Nokia), and South Korean carriers all publishing competing architectural positions ahead of ITU's IMT-2030 finalisation. The AI-native framing etisalat and Khalifa University are advancing is consistent with 3GPP's Release 18 "AI/ML for NR" work items, but the question of which architecture becomes the reference implementation remains open. For enterprise buyers and infrastructure architects, the practical significance of this research lies in its potential to influence which AI integration patterns are baked into commercial network equipment from 2029 onwards, with direct consequences for how operators provision and manage private 5G/6G slices for industrial and smart-city applications.