6G news and analysis for the week ending Friday, 14th August, 2026.
Welcome to another week of 6G readiness news. NTT Docomo and Samsung Electronics lead our coverage for this week with the validation of AI-RAN optimisation applied to individual users rather than to the cell as a whole. The headline result is an incremental improvement but there is far more interesting detail contained within their news release. The two companies have already taken their data collection method into 3GPP and intend to contribute the underlying prediction technology as well.
South Korea provides the week’s policy story. The Ministry of Science and ICT has begun rebuilding the way it calculates spectrum reallocation fees. A newly convened research group has been asked to move the methodology beyond past revenue figures and auction prices. Whatever framework emerges will likely set the cost of spectrum for SK Telecom, KT and LG Uplus.
Meanwhile, du and Khalifa University have published a 6G blueprint arguing that operators should reposition as providers of intelligence rather than connectivity. It’s a vision document and the premise is rather familiar. The upper 6 GHz field data contained inside the paper is more useful and it adds to an evidence base that regulators will find useful in the run up to WRC-27.
Also this week, VIAVI Solutions launched a channel emulator for 6G and Wi-Fi 7/8 testing, with a waveform claim that deserves scrutiny, and LMT joined Europe’s SNS JU Advisory Group. We also have our four selected research papers, which this week cover spectrum coexistence, digital twin networks, antenna orientation and semantic sensing. What a treat...
NTT Docomo and Samsung Electronics have validated AI-RAN optimisation applied at the level of the individual user rather than uniformly across a cell. The technology learns each user’s movement patterns and service usage, then predicts service quality degradation before it occurs. In testing conducted in Japan in January 2026, using simulations built on data from Docomo’s commercial network and a local 5G trial field, the frequency of communication speed degradation fell from 13.1 per cent to 7.2 per cent. The announcement was published by Docomo on 10 August. Masafumi Masuda, Senior Vice President and General Manager of the 6G-Tech Department at NTT Docomo, said the work represents “a significant step toward realising more adaptive and user-centric communication experiences in future networks.”
What’s new: The genuinely new element is the data collection method rather than the optimisation itself. Rather than bulk-collecting radio environment data, the system selectively gathers only the information relevant to a specific user problem, reducing network load. The companies jointly contributed this approach to 3GPP discussions in February.
Why this matters: Samsung and KT validated a closely comparable per-user AI-RAN concept on a live commercial network with around 18,000 users, which we covered in our 19th December edition. This latest result uses simulation built on collected network data, so on evidentiary weight it sits behind that December work despite arriving eight months later. What it adds is a quantified figure for a specific impairment and a second tier-one operator partner for Samsung. Both companies frame the work as customer-experience-driven rather than focused on radio link failure prediction. The MDT mechanism they rely on, Minimisation of Drive Tests, is an existing 3GPP feature for anonymised quality reporting from handsets, so the innovation is in what gets collected rather than in new instrumentation.
Implications for 6G standards: Data collection is where AI-native RAN ambitions come up against practical constraints, because models cannot be trained on data that networks cannot afford to gather. A contribution on selective collection methods addresses a real bottleneck in the Release 20 AI/ML study items.
View source material to read more
South Korea’s Ministry of Science and ICT has begun a comprehensive overhaul of how it calculates spectrum reallocation fees, framing the work as preparation for the AI and 6G era. The ministry announced on 13 August that it has convened a research group of experts in law, economics, management and technology, which has held its inaugural meeting. The group will examine the economic value of spectrum, shifts in the market and policy environment, and international practice. The central question is whether the methodology should move beyond past revenue figures and auction prices to incorporate spectrum usage efficiency, carrier investment incentives and competitiveness. No timeline for reform proposals has been disclosed. Note that the report we have seen summarises a Ministry of Science and ICT announcement rather than reproducing it.
What’s new: The significant element is the proposed shift in valuation basis. Reallocation fees in Korea have been calculated with reference to historical revenue or prior auction outcomes, both of which are backward-looking measures. Incorporating usage efficiency and investment incentives would make the fee a policy instrument rather than simply a charge.
Why this matters: This is an interesting follow-up to the news we reported on 29th May, when domestic experts questioned whether Korean operators would fund next-generation investment at all. Combined ARPU across SK Telecom, KT and LG Uplus has been falling since 2022, and the government has already made 5G Standalone deployment a condition of reallocating 3G and LTE spectrum. Spectrum cost is therefore no longer a separable question from 6G readiness, and the ministry appears to recognise this. The 2018 allocation of 28 GHz spectrum, which the carriers later handed back on profitability grounds, remains the cautionary precedent. Few administrations have attempted to build investment incentives directly into spectrum pricing, so the outcome of this work will be watched closely by many other regulators.
Implications for 6G standards: Fee methodology does not touch 3GPP specifications directly, but it shapes which bands operators are willing to acquire and clear. Korea’s positions at WRC-27 and its contributions through TTA will reflect what its carriers can afford to deploy.
View source material to read more
UAE operator du has published a white paper with Khalifa University titled “The Intelligence Network: A 6G Blueprint for Telcos.” The paper argues that connectivity alone no longer generates proportional economic value, and that operators should evolve into providers of intelligent infrastructure. It sets out an AI-native architecture built on a unified data fabric, AI/ML lifecycle management, cloud-native network functions, AI agents, digital twins and telecom world models. The most substantive technical content concerns spectrum. du reports that field evidence indicates upper 6 GHz can support macro-cellular deployment with coverage behaviour comparable to 3.6 GHz under equivalent conditions, with 200 MHz channels opening a path to multi-gigabit performance.
What’s new: The paper’s practical contribution is the field data rather than the architecture. It also makes an explicit case for beginning the autonomy journey during 5G-Advanced rather than at 6G launch, naming RAN energy optimisation, proactive fault management and enterprise SLA management as the scenarios where autonomy carries clear value and manageable risk.
Why this matters: The upper 6 GHz finding is consistent with independent results from Telefónica and Nokia at the Polytechnic University of Valencia, which we covered on 14th November, and from SoftBank and Nokia in Tokyo’s Ginza district. Three separate operators now report comparable coverage between upper 6 GHz and existing mid-band 5G using existing sites. That consistency matters more than any single trial, particularly for regulators weighing the licensed mobile case against Wi-Fi claims on the same band. The architectural material overlaps considerably with the e& and Khalifa University paper we covered on 20th March, which proposed a dedicated AI plane. Note that du does not publish trial parameters, site counts or measurement methodology here, so the coverage claim is a summary rather than a verifiable result. The monetisation chapter, covering AI-as-a-Service, Agent-as-a-Service and SLA-as-a-Product, is vision material rather than strategy.
Implications for 6G standards: du is a member of the UAE upper 6 GHz commercial ecosystem launched at the SAMENA Council summit, and the band is designated n104 by 3GPP. Accumulating operator field evidence in this range strengthens the case for early FR3 specification work under Release 21.
View source material to read more
VIAVI Solutions has launched Vertex 6.0, the latest version of its channel emulation platform, offering 400 MHz instantaneous bandwidth and support for carrier frequencies up to 23.6 GHz. Channel emulators recreate real-world radio propagation conditions in the laboratory, allowing equipment to be tested against realistic signal behaviour before field deployment. The platform supports FR3 bands with up to 1 GHz of bandwidth, multi-user MIMO with phase alignment, and modes covering non-terrestrial networks across low, medium and geostationary orbits. It is the first Vertex upgrade since VIAVI acquired Spirent Communications’ channel emulation business.
What’s new: The FR3 capability is the element most relevant to 6G, alongside integration with VIAVI’s FR3 MIMO converter and ray tracing for integrated sensing and communication, or ISAC. Each chassis supports 36 RF ports and up to 1.6 GHz of bandwidth.
Why this matters: Let’s face it, test and measurement capability is unglamorous and consistently underestimated. Yet equipment vendors cannot validate FR3 designs without emulation hardware that operates at those frequencies, and the upper mid-band is where most early 6G deployment is now expected. The “industry’s first” framing should be read carefully though. Rohde & Schwarz supplied vector signal generation and channel emulation for the Nokia Bell Labs AI receiver work we covered on 7th November 2025, and Keysight provides user equipment emulation within the KDDI and NVIDIA digital twin.
Implications for 6G standards: Conformance testing sits with RAN4 and RAN5, and RAN5 does not define requirements independently but inherits them from RAN4. Emulation platforms will eventually need to track whatever waveform and numerology decisions emerge, rather than anticipate them.
View source material to read more
Latvian operator LMT has confirmed it is one of 16 organisations joining the Smart Networks and Services Joint Undertaking Advisory Group, the body providing strategic input into the European Union’s digital connectivity priorities. Members contribute to the SNS JU annual work programmes, the Strategic Research and Innovation Agenda, and the research framework for the 2028 to 2034 EU budget cycle. LMT said it will play a leading role in two SNS JU initiatives. Shield 6G addresses resilience and protection against cyber threats in next-generation networks, while IoT Zero targets zero-power and ultra-low-power device and network architectures. The appointment was reported by Telecompaper.
What’s new: The interesting news is the budget cycle rather than the appointment itself. Advisory Group members shape the research framework for the funding period that will run through the entire 6G standardisation and early deployment window.
Why this matters: Both named projects have appeared in these pages already. We first covered SHIELD-6G on 12th June, when University College Dublin was named as lead of the eight million euro project. IoT-ZERO featured among the four SNS JU grants secured by IMDEA Networks that we reported on 5th June. LMT is a small operator by European standards, but it has built a disproportionate profile in defence-adjacent connectivity and drone detection work, which aligns with the resilience agenda now dominating European 6G research.
Implications for 6G standards: SNS JU projects feed concepts into 3GPP and ETSI through their industrial participants rather than directly. Resilience and ultra-low-power IoT both map to active themes in the Release 20 and 21 study programmes.
View source material to read more
Researchers from Northeastern University propose a framework in which cellular communications, radio frequency sensing, radionavigation and radar share a common pool of physical resource blocks under centralised coordination. The work formulates the problem as constrained optimisation across scheduling, resource assignment and power allocation, with separate quality of service constraints for each service type. Testing uses site-specific ray tracing on the BostonTwin urban digital twin.
Why this matters: Regulators across the US and Europe are under pressure to repurpose federal sub-6 GHz spectrum without displacing incumbent users, a tension running through the “Winning the 6G Race” memorandum and its associated band studies. This paper treats coexistence as a coordination problem rather than an exclusion problem, which is the direction policy is heading. The inclusion of positioning accuracy constraints alongside sensing and throughput is notable, given how many 6G proposals assume sensing can be added without cost to capacity.
View source material to read more
Charalampos Oikonomidis and colleagues categorise digital twin network architectures into passive monitoring twins and active control twins, then assess the enabling technologies behind each, including ray tracing, reconfigurable intelligent surfaces and edge computing. The survey’s distinguishing contribution is a computational complexity analysis intended to test hardware scalability and identify inference bottlenecks. It closes on unresolved problems including GPU limitations, actuation latency and security.
Why this matters: Digital twins ranked among the top requests at the 3GPP 6G Workshop and have since attracted heavy vendor investment from Nokia, NVIDIA and KDDI. Most published work remains conceptual, and this paper is unusual in asking whether the proposed architectures are computationally affordable. The passive and active distinction is also useful, because a twin that only observes carries very different risk from one that acts on the live network.
View source material to read more
Chen Xu and colleagues develop an electromagnetic channel model for spatially reconfigurable antenna systems, which adjust the position and orientation of individual antenna elements. Conventional models treat antennas as ideal isotropic radiators and ignore the vector nature of electromagnetic propagation, which the authors argue leaves performance on the table. Full-wave simulations and physical measurements support the model, and an optimisation method built on it improves sum rate by up to 16.6 per cent over movable antennas.
Why this matters: Reconfigurable antenna research is now crowded, and we surveyed the wider field on 19th June. The distinctive finding here is that orientation influences achievable rate more than displacement does, which cuts against the emphasis of much movable antenna work. If it holds, it changes what the hardware needs to do. The inclusion of experimental measurement rather than simulation alone gives the result more weight than most papers in this area.
View source material to read more
Haotian Wang and colleagues propose a framework that jointly optimises source and channel coding rates alongside transmit beamforming for bi-static integrated sensing and semantic communication systems. Bi-static here means transmitter and receiver sit at separate locations, which introduces synchronisation error that the authors model explicitly. The paper derives a bound on target positioning accuracy under imperfect synchronisation and reports gains over established benchmarks.
Why this matters: Most work combining sensing with semantic communication treats compression of sensing data as the problem to solve, leaving the two functions loosely coupled. Designing the transmission framework for both tasks together is the harder and more useful question for 6G, where ISAC and AI-driven communication are expected to share the same radio resources. Treating synchronisation error as a design constraint rather than an assumption is realistic, since separated transceivers are the likely deployment case.
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.