6G Technologies and Trends From Vision to Practical Testbeds

Last week, we looked at the visible and invisible technologies that will power future 6G networks. The visible side includes things that are easy to describe, such as immersive services, digital twins, integrated sensing, non-terrestrial connectivity and new spectrum. The invisible side is just as important, because many of the real changes will happen inside the network through cloud-native architecture, software-defined infrastructure, automation, AI-native optimisation and new approaches to testing.

This week’s post continues that theme by looking at a lecture by Sarah LaSelva, Chief Product Marketing Manager for NI’s RF portfolio, now part of Emerson’s Test & Measurement business. The talk, titled 6G Technologies and Trends, was delivered at the IEEE Central Texas Consultants Network and Life Member Affinity Group joint meeting on 2 April 2026. It provided a useful snapshot of where the 6G discussion stands today, not just from a standards and technology perspective, but also from the point of view of validation, testing and practical implementation.

One of the most useful parts of the talk was the reminder that 6G is not appearing in isolation. 3GPP held its first dedicated 6G workshop in March 2025, and the industry is now working through Release 20 as the study phase for 6G. Release 21 is expected to become the first normative release containing the initial set of 6G specifications, with commercial deployment broadly associated with the 2030 timeframe. At the same time, 5G-Advanced continues, and even LTE-related work has not completely disappeared. Mobile generations overlap for a long time, so 6G should be seen as an evolution layered on top of a very active 4G, 5G and 5G-Advanced ecosystem rather than a clean break from everything that came before.

Sarah highlighted several areas where 5G-Advanced and early 6G studies are already touching each other. These include non-terrestrial networks, GNSS-resilient operation, integrated sensing and communication, and AI/ML. The AI/ML discussion was particularly interesting because it moved beyond the simple statement that AI will be everywhere. One practical issue is that the handset and the base station may not be trained using the same data. Operators, vendors and device companies all have valuable proprietary datasets, and there are obvious commercial, privacy and competitive reasons why they may not want to share them. This leads to the idea of dual-sided models, where the device and network may each use AI, but not necessarily trained in the same way.

That point is important because AI-native 6G is sometimes presented as if the industry will simply add a common intelligence layer and everything will optimise itself. The reality is more complicated. AI in the RAN will have to work across different vendors, datasets, policies, hardware capabilities and timing constraints. It will also need to be tested and trusted. In a mobile network, a model that performs well in simulation may still fail if it cannot run within the timing budget of the physical layer or MAC layer.

Spectrum was another major part of the lecture. Every mobile generation has benefited from access to more bandwidth, but finding clean spectrum for 6G is becoming harder. The early excitement around sub-THz frequencies has not disappeared, but the more practical near-term focus appears to be the upper mid-band, often discussed as FR3, roughly in the 7 to 24 GHz range. Sarah pointed to the 7.125 to 8.4 GHz range as one of the most promising areas being discussed, sometimes described by others as a kind of golden band because it could offer a useful compromise between bandwidth and propagation.

This is where the lessons from 5G mmWave matter. Millimetre wave brought large bandwidths but also major deployment challenges. Outdoor-to-indoor penetration, blockage, multipath complexity and economics all limited its impact in many markets. For 6G, the industry seems to be looking for bands that offer more capacity than today’s sub-6 GHz spectrum but with better propagation and more practical deployment characteristics than the highest mmWave and sub-THz bands. The challenge is that these frequencies are already used by other services, including satellite systems, which means sharing, coexistence and regulatory work will be central to the 6G story.

The talk also brought out the business reality behind the technology. Operators spent heavily on 5G deployments, but many did not see the new revenue streams they had hoped for. This is why cost reduction and new revenue opportunities are now central to 6G discussions. Cost reduction is linked to software-defined networks, open interfaces, energy efficiency and the possibility of avoiding expensive hardware replacement cycles. New revenue opportunities are often linked to non-terrestrial networks, integrated sensing and communication, network APIs, digital twins and new enterprise services.

Energy efficiency was treated not as a nice-to-have feature, but as a core design driver. Radio access networks consume significant power, with base stations, power amplifiers, cooling systems and always-on operation contributing heavily to energy use. A simple but important point is that the network cannot just switch things off blindly, because users still expect coverage, capacity and reliability. Energy optimisation therefore requires a careful relationship between traffic demand, radio configuration, quality of service and power consumption.

This is where AI may have a useful role, but again the talk avoided the trap of assuming AI is always the answer. AI can help with channel estimation, beam management, energy optimisation, flow management and RF impairment compensation, but conventional signal processing remains extremely strong in many parts of the wireless chain. In some areas, existing DSP techniques are already close to theoretical performance limits. The lesson is not that AI should replace everything, but that it should be used where it provides a measurable advantage.

One of the most interesting examples from NI’s research was the neural receiver. In a conventional receiver chain, functions such as channel estimation, equalisation, symbol detection and channel decoding are handled using well-established signal processing blocks. A neural receiver tries to combine some of these tasks using a machine learning model. That sounds attractive, but the practical testbed results showed the importance of implementation realism. A model that looks excellent in simulation may need to be reduced in size to meet real-time constraints, and once that happens the gains may reduce. The performance also depends on what the model was trained for, which means a general model may not always deliver the same gains as a model trained for a specific scenario.

This is a very useful message for 6G research. We need not only better algorithms, but also better ways to test algorithms under realistic conditions. Timing, hardware constraints, RF impairments, channel models, power consumption and unexpected corner cases all matter. This becomes even more important when AI models are embedded inside real network equipment. Traditional conformance and performance testing may not be enough, because AI-enabled systems can behave differently depending on the scenario, the training data and the environment.

Integrated sensing and communication, or ISAC, was another major theme. ISAC is one of the most visible 6G candidate capabilities because it changes the role of the mobile network. Instead of only transmitting and receiving data, the network could also sense the physical environment. This could support applications such as digital twins, traffic monitoring, localisation, industrial safety, gesture recognition, factory automation and situational awareness.

The lecture made a useful distinction between different levels of integration. The simplest approach is to co-locate communication and sensing systems at the same site. A deeper level is to share spectrum or hardware. The most ambitious approach is to share the same signal for both communication and sensing. In practice, the industry may not reach full signal sharing in all cases, because sensing and communication have different requirements. Radar-like sensing may benefit from frequent or continuous transmission, while energy-efficient communication networks often want to sleep, mute or reduce transmission when demand is low.

Sarah also explained why OFDM remains attractive for ISAC. There has been a lot of discussion around whether 6G needs a completely new waveform, but OFDM already has useful properties for sensing. Reference signals and subcarriers can be used to estimate motion, range and other characteristics. However, the trade-offs matter. A waveform optimised for high data throughput may not be ideal for sensing performance. The hardware also matters, because a radar receiver and a smartphone receiver are designed with different cost, sensitivity and performance assumptions.

The NTN discussion was equally practical. Satellite connectivity is now much more realistic because launch costs have fallen and LEO constellations are being deployed at scale. We are already seeing early direct-to-device and satellite messaging capabilities, although data rates and service capabilities remain limited. For 6G, the ambition is broader integration between terrestrial and non-terrestrial networks, potentially supporting global coverage, emergency connectivity, remote areas, maritime, aviation and IoT applications.

At the same time, NTN is not a solved problem. Different architectures are possible, including regenerative and transparent payloads, satellite-based base station functions, ground-based gateways and different waveform choices. Mobility, Doppler, delay, spectrum access, link budget, antenna constraints and device compatibility all add complexity. There is also the business and sustainability question of how frequently LEO satellites need to be replaced and whether the economics remain attractive over the long term. These questions are just as important as the radio technology.

The final part of the talk looked at the softwarization of networks. This links strongly to the invisible technologies discussed in last week’s post. The industry has been moving from dedicated hardware towards centralised RAN, virtualised RAN, cloud RAN and Open RAN. Open RAN introduced the ambition of disaggregated components, open interfaces and multi-vendor interoperability. The reality has been more difficult than the vision, but the direction of travel remains important for 6G.

A key technical point is the functional split between the radio unit and the distributed or centralised processing. With O-RAN split 7.2, some lower physical layer processing is moved into the radio unit, including FFT processing. This can help standardise interfaces and reduce fronthaul requirements, but it can also constrain researchers who want full flexibility to experiment with new waveforms, frequency ranges or physical layer designs. This is one of the tensions in 6G research: commercial systems need stable, interoperable and cost-effective interfaces, while research often needs flexibility and experimentation.

Sarah also mentioned the growing interest in open-source RAN software stacks and government-backed initiatives such as OpenCU/DU concepts. The broader idea is to move towards a more transparent, programmable and software-driven RAN, where innovation can happen faster than the traditional 10-year mobile generation cycle. This could support security transparency, supply chain visibility and faster service development, but it also raises difficult questions about carrier-grade reliability, vendor business models, integration effort and long-term support.

For me, the most important takeaway from the lecture is that 6G is becoming less of a vision exercise and more of an engineering problem. The visible themes are still there: AI-native networks, ISAC, NTN, FR3 spectrum, digital twins and new services. But the hard work is increasingly in the invisible layers: testbeds, datasets, power measurements, timing constraints, software architecture, interfaces, model validation, energy profiling and practical deployment economics.

This is why test and measurement companies are so important in the 6G ecosystem. As soon as we move from PowerPoint visions to real hardware and software, we need to know what is actually happening. Does the AI model still work when it runs in real time? Does the energy saving feature reduce power without damaging user experience? Does an ISAC waveform still communicate reliably while sensing the environment? Does an NTN link work with ordinary devices in realistic propagation conditions? Can an open RAN implementation be made secure, reliable and carrier-grade?

The answer to these questions will determine whether 6G becomes a meaningful evolution or simply a collection of attractive research topics. The lecture is therefore worth watching because it connects the big 6G themes with the practical realities of experimentation and validation. It also reinforces the point from last week’s post: the future of 6G will be shaped not only by the technologies users can see, but also by the hidden engineering work that makes those technologies reliable, efficient and deployable.

The talk is embedded below:

Related Posts: 

Posted by on

Comments

Post a Comment