Stephen Douglas, head of strategic thought leadership at Keysight Technologies, said during the recent RCR’s 6G Forum that future 6G technology will be standalone from the beginning with its own dedicated 6G core
In sum – what to know
Timeline tightens – Keysight’s Stephen Douglas said early demonstrations could take place around the 2028 Los Angeles Olympics, with potential first commercial systems in mid-to-late 2029.
AI design debated – 3GPP is studying whether AI should be integrated into network functions, supported through a separate AI domain or enabled through a more agent-based architecture.
Consensus remains essential – Docomo Euro-Labs’ Bahador Bakhshi said competing approaches to 6G architecture remain under discussion, with a compromise expected to emerge through further work in the coming months.
The 6G core is progressing toward key standardization milestones, but the industry has yet to agree on how artificial intelligence should be incorporated into the architecture, according to Stephen Douglas, head of strategic thought leadership at Keysight Technologies, and Bahador Bakhshi, senior network architect at Docomo Euro-Labs.
Speaking during RCR Wireless News’ 6G Forum 2026, Douglas said the timelines toward possible commercial systems in 2029 had become compressed. Early demonstrations could take place around the 2028 Los Angeles Olympics, while potential commercial systems could follow in mid-to-late 2029.
Reaching that schedule requires mature, interoperable standards, prioritized market-demand use cases, functioning chipsets and devices, and an agreed 6G core architecture. Douglas said one lesson from 5G was the time taken to move from non-standalone deployments anchored in 4G to standalone networks. For 6G, Douglas said that the plan is that the techology will be standalone on day one with its own dedicated 6G core.
The new core is expected to support capabilities that were not central to the 5G core’s original design, including AI-native control, integrated sensing and communications (ISAC), tighter integration with computing, and native support for non-terrestrial networks. Douglas also highlighted scaling challenges in signaling and service-based architecture, alongside security requirements involving post-quantum cryptography and zero-trust principles.
However, the architecture remains unsettled. “The 6G core architecture and the network functions are not yet agreed,” Douglas said, adding that many proposed functions remain under study rather than forming part of a settled design.
3GPP studies different AI approaches
Bakhshi explained that 3GPP is not standardizing AI models or learning algorithms themselves. Instead, its role is to define the architecture, components, interfaces, and procedures that allow AI to operate within real networks.
The work builds on AI capabilities already present in 5G, where a centralized analytics function collects data from network components and provides predictions that other functions can use for decisions, such as user mobility management. Bakhshi said this approach has limitations, including centralized AI processing, limited operator control and observability, and restricted support for external AI-based functions.
For 6G, the study includes intent-driven networking, AI capabilities in additional network functions, closed-loop control, monitoring and control of AI components, and interoperability and roaming. Bakhshi said the work spans 22 solution variants.
Two broad architectural approaches have emerged. One favors a more revolutionary, multi-agent architecture in which agents could handle flexible user requirements and determine procedures dynamically. The other favors an evolutionary approach, enhancing the 5G architecture through additional functions while reusing existing procedures and service invocations.
Bakhshi said companies differ in how much AI should be integrated into the network and how extensively the architecture should change. He also identified concerns about AI reliability and sustainability as factors complicating the discussions.
He does not expect the competing approaches to be resolved immediately. Instead, he said, further discussions in the coming months are likely to produce a compromise that participants can agree on through the 3GPP consensus process.
New functions depend on justified needs
The speakers also discussed how the industry should decide which capabilities require new 6G functions rather than adaptations of existing 5G ones.
Douglas said both technical necessity and business requirements can justify new functions. He cited sensing and AI communications as examples where current 5G core capabilities may be insufficient.
Potential functions under study include combining the user plane function with edge computing, making the access and mobility management function more AI-native and intent-driven, and introducing functions for sensing operations, AI-agent discovery and communication, and large-scale data orchestration.
Douglas stressed that these remain proposals rather than agreed specifications. He also said operators will need mechanisms to keep autonomous network operations within defined limits.
Table of Contents
