Much of the discussion around 6G naturally concentrates on networks: new spectrum, new radio technologies, AI-native architectures, sensing, distributed computing and the eventual evolution of the 5G system. At Mobile Korea 2025, BK Kim, Head of Modem R&D at MediaTek, looked at the journey from Beyond 5G towards 6G from a different direction: what does all of this mean for the device?
That is an important question. Whatever capabilities eventually appear in 6G networks have to be implemented in smartphones, vehicles, wearables, robots and other devices without making them excessively expensive, power hungry or complex.
MediaTek summarised its 6G device vision around three broad themes: sustainable and green, intelligent and AI-based, and available everywhere.
There was also an interesting change in emphasis from simply moving more bits. Kim suggested that future communication increasingly needs to support the delivery and interpretation of meaning rather than just conventional data transfer. This fits naturally with the emergence of multimodal and agentic AI, where networks increasingly carry prompts, context, sensor information and machine-generated data between devices and computing resources.
MediaTek's development roadmap presented at the event was quite ambitious. Kim described an “AI first” phase, followed by AI-related demonstrations between the UE and network, a real-device demonstration targeted around MWC 2027, and the possibility of showing pre-commercial 6G devices, robots or other new device types around the 2028 Olympic timeframe. This should be treated as an R&D ambition rather than a commitment to commercial 6G services in 2028, but it demonstrates how early device vendors need to start developing the underlying technologies.
One of the most useful parts of the presentation was the discussion of lessons from 5G.
Kim argued that 5G has not always delivered the coverage economics expected when compared with LTE. Higher-frequency deployments can require more infrastructure, while coverage and deployment cost remain significant challenges in many markets. From MediaTek's perspective, 6G therefore cannot simply offer more capacity. It also has to improve the practical economics and coverage of mobile broadband.
This places considerable responsibility on the device.
MediaTek is investigating techniques including higher UE transmit power, improved beamforming, antenna techniques and better channel estimation to extend coverage from the terminal side. At the same time, wider spectrum allocations and greater bandwidth will create their own RF, power and thermal challenges.
Energy efficiency is particularly important. If AI applications, sensing and much higher data generation significantly increase traffic, devices cannot simply increase their processing and radio activity proportionally. Battery life, heat dissipation and RF power efficiency therefore become part of the 6G system design rather than merely implementation details.
An interesting example discussed in the presentation was Sub-Band Full Duplex, or SBFD.
Instead of a conventional TDD system separating uplink and downlink transmissions completely in time, SBFD allows uplink and downlink operation within different portions of the same TDD carrier at the same time. This can improve uplink availability and latency, but creates a difficult self-interference problem, particularly inside a small device where transmitting and receiving antennas are physically close together. Kim noted that MediaTek had already investigated feasibility in its chipset work.
MediaTek subsequently quantified some of these benefits in its MWC technology demonstrations. Its SBFD implementation was presented as capable of delivering 43% greater uplink coverage, twice the network capacity and around half the latency, while remaining feasible in a smartphone form factor.
This is a good example of why the device perspective matters. A radio technique can look extremely attractive at system level, but it only becomes useful if the RF front end, antennas, interference cancellation, modem processing, power consumption and thermal characteristics can all be made practical in a real product.
Since the Mobile Korea presentation, MediaTek has provided a clearer picture of how some of these ideas are progressing.
At MWC 2026, the company demonstrated what it described as the world's first 6G radio interoperability using its MediaTek 6G Test Platform. The prototype was positioned around what MediaTek calls 6G for Physical AI, where robots and other devices can access edge computing resources when local processing is insufficient. The radio is designed to vary the balance between throughput, latency, robustness and power consumption depending on the AI service being supported.
MediaTek also worked with Ericsson on a 6G centimetre-wave prototype data call. The demonstration integrated an Ericsson 6G RAN prototype with a MediaTek UE prototype, including the radio and baseband processing up to the IP layer. It also demonstrated contention-based buffer status reporting, an anticipated 6G feature intended to reduce latency at scale.
This is significant because it moves part of MediaTek's 6G activity beyond isolated component experiments towards early interoperability between a device-side platform and network equipment.
AI has also moved much more directly into the radio.
At MWC 2026, MediaTek demonstrated AI-accelerated uplink transmit diversity developed with Keysight and vivo. Instead of relying entirely on predefined rules for selecting transmission behaviour, an AI model can adapt to changing radio conditions. MediaTek reported up to a 160% improvement in cell-edge uplink throughput compared with its traditional rule-based transmit-diversity approach.
Potentially more important than the headline performance figure is how the AI model is maintained. The MediaTek and Keysight prototype included site-specific model retraining and over-the-air model updates, recognising that an AI model trained for one radio environment may not work equally well everywhere.
This starts to make the idea of an AI-native air interface more tangible. AI is not simply being used somewhere in network management. It can become part of how the UE itself adapts radio transmission to its environment.
Another interesting development is MediaTek's 6G Device Cloud concept.
MediaTek describes an Intelligent Personal Device Cloud in which CPE, PCs and mobile devices can collaborate through secure context sharing, cross-device orchestration and unified AI computing. Rather than treating every personal device as an isolated computing endpoint, AI workloads and context can potentially be coordinated across the collection of devices. MediaTek also envisages this personal device environment extending towards edge and cloud computing when additional resources are required.
Related to this, but addressing the radio rather than the computing side, is Cross-Device Collaborative MIMO, or Co-MIMO. MediaTek describes Co-MIMO as aggregating antenna resources across multiple devices, giving the example of a smartphone and a car, to improve performance at the cell edge.
MediaTek developed the idea further at Computex 2026. Its 6G Device-Cloud prototype, powered by the MediaTek M90 and a 6G Co-MIMO module, uses shared antennas and boosted power to allow devices to collaborate. At Computex 2026, MediaTek reported more than 60% improvement in downlink throughput from its Co-MIMO demonstration, with smartphones or wearables able to aggregate signals from other indoor 6G devices.
This broadens the conventional model of the UE as an independent endpoint. Rather than being simply:
Device → Base Station → Network
There are therefore two related forms of device collaboration:
Personal Device Cloud: Multiple Devices ↔ Context Sharing / Collaborative AI Computing ↔ Edge / Cloud
Co-MIMO: Multiple 6G Devices → Aggregated Antenna Resources → 6G RAN
The first is primarily about AI, context and computing, while the second is about radio cooperation between devices.
MediaTek refers to the broader computing concept as an Intelligent Personal Device Cloud, where context and AI computing can be coordinated across personal devices. Co-MIMO complements this on the connectivity side by allowing multiple devices to cooperate at the radio level.
The RF hardware required to make future spectrum practical is progressing as well. MediaTek and Skyworks demonstrated an early 6G FR3 RF front-end reference design covering approximately 6.425 GHz to above 7 GHz at MWC 2026. The collaboration is notable because moving into wider channels and higher frequencies is not simply a modem problem. Power amplifiers, filters, antennas, thermal performance and the rest of the RF front end all have to evolve at the same time.
Taken together, these developments make the Mobile Korea presentation useful in retrospect.
MediaTek's device-centric view of 6G is not primarily about building a smartphone capable of displaying an enormous peak data-rate number. The bigger challenge is creating devices that can maintain useful coverage, operate efficiently across wider spectrum, use AI within the radio, cooperate with other devices, access distributed computing resources and do all of this without unacceptable battery, thermal or cost penalties.
The company's 2026 demonstrations also suggest an evolution from the familiar idea of the UE as an endpoint towards the UE as an intelligent participant in a distributed communication and computing system.
There is still a long distance between these prototypes and commercial 6G. Technologies such as collaborative MIMO, AI-native radio optimisation and centimetre-wave communications still have to pass through research, standardisation, ecosystem development and eventually highly constrained commercial device implementations.
I have also not seen a newer MediaTek announcement turning the 2028 pre-commercial device ambition mentioned at Mobile Korea into a formal product or commercial launch commitment. It is better understood as an indication of the company's development trajectory.
Nevertheless, the progression from the Mobile Korea 2025 roadmap to the MWC and Computex demonstrations in 2026 is already visible.
For anyone following 6G development, MediaTek's perspective provides a useful reminder: a new generation of mobile technology only becomes real when all of those ambitious network capabilities can actually be made to work inside practical devices.
Related Posts:
- Free 6G Training: 6G News and Announcements from MWC 2025
- Free 6G Training: MediaTek’s View on Enabling a Healthy and User-Centric 6G Device Ecosystem
- Free 6G Training: MediaTek 6G Satellite and Terrestrial Network Convergence Technology White Paper
- Connectivity Technology Blog: 5G NR-NTN Demos make a Debut at MWC 2023
- Free 6G Training: MediaTek Releases 6G Vision White Paper
- The 3G4G Blog: 5G Dynamic Spectrum Sharing (DSS)



Comments
Post a Comment