Panelists favor integration over a single 6G architecture
In sum – what we know:
- Service over bandwidth – Autonomous systems may need reliable kilobits per second rather than gigabits, so different applications should drive different KPIs.
- Integration over architecture – Legacy infrastructure rules out a single common 6G design, leaving common interfaces, orchestration and AI as the practical path.
- Unsettled economics – No killer 6G application has emerged, and the business case for covering remote areas still hinges on users willing to pay.
What does ubiquitous connectivity actually mean? That was the focus of a panel at the recent 6G Forum, where representatives from Northeastern University, ETSI, University College London, and Battelle, covered the gap between the aspiration — seamless service regardless of where a user happens to be — and the engineering reality of stitching together terrestrial, satellite, and aerial networks that were never designed to work as one.
The panelists largely agreed on two things. First, ubiquity doesn’t mean broadband everywhere. It means uninterrupted service matched to the application, whether that’s gigabits for a consumer or reliable kilobits for an autonomous system. Second, the path forward runs through integrating heterogeneous networks rather than replacing them with a single common architecture. Sub-terahertz frequencies and integrated sensing came up as tools that could help fill coverage gaps and improve handover — though the business case for remote-area connectivity remains far from settled.
Ubiquity means service on demand, not coverage everywhere
The term “ubiquitous connectivity” gets thrown around casually in 6G circles, but the panelists were quick to draw a line between the marketing shorthand and the actual objective. Christos Masouros, professor of signal processing and wireless communications at University College London, framed it plainly. “Different applications will require different types of connectivity, or different KPIs become important, whether it’s the capacity or data rate, or whether it’s going to be latency, reliability,” he said. “An autonomous system, for example, might not need gigabits per second, but it might need very reliable bits or kilobits per second. So ubiquitous connectivity is more about service on demand, not broadband everywhere. That’s not what we need. That would lead to over-engineering.”
Josep Miquel Jornet, associate dean of research at Northeastern University, echoed that view but tied it to geography. In a dense urban area with strong base station coverage, non-terrestrial links add little. The value emerges as users leave cities, moving into areas where deploying fixed infrastructure doesn’t make financial sense. That’s the gap ubiquity is meant to close — not blanket broadband, but the right connection when the preferred path isn’t available.
Sam Roberts, spectrum solutions portfolio manager at Battelle, brought a defense lens to the conversation. For military users, the priority isn’t necessarily coverage for its own sake — it’s continuity. Can essential communications survive jamming, interference, congestion, or the loss of a preferred link? In that context, resiliency becomes the defining requirement. Roberts noted that the system needs to understand signal priority, where a command-and-control message from a commanding officer outranks bulk drone data, but those priorities can be flipped on demand when a high-priority mission requires raw video footage from that same drone. “It’s about can you continue to operate when your preferred path degrades or disappears,” he said.
Jussi Kahtava, a technical officer at ETSI, broadened the scope beyond consumers entirely. 5G already introduced verticals, and 6G will expand that further. Connectivity covers not just terrestrial and non-terrestrial networks but also IoT, private networks, and the seamless handover between private and public infrastructure. “There are many, many aspects,” Kahtava said. “But yes, that’s where we are moving.”
Integrating satellite and terrestrial networks
One of the more grounded exchanges in the panel focused on the gap between how 5G handles non-terrestrial networksand what 6G aims to do differently. In 5G, terrestrial networks operate on IMT bands while satellite NR runs on S-band and Ku-band — separate spectrum, loosely coordinated. Dual connectivity exists in 3GPP standards, where a session could theoretically run through both a terrestrial base station and a satellite link, but as Kahtava pointed out, it isn’t used in practice.
6G is supposed to change that. The goal is a unified air interface at the physical layer, with integration handled at the architecture level rather than bolted on after the fact. Kahtava flagged a concrete technical challenge that’s already surfacing in standards work. Some satellite operators, including AST SpaceMobile and SpaceX, want unmodified legacy devices — even older LTE handsets — to connect directly to satellites. Those devices don’t know they’re talking to a satellite; they think it’s another base station. Meanwhile, Release 17–era terminals do their own pre-compensation for Doppler and timing. When both types of terminal operate on the same spectrum, which is the direction 6G is heading, scheduling and interference problems emerge because signals arrive at the satellite receiver at different timings regardless of what the network does. A new work item addressing exactly this was started at a recent 3GPP RAN plenary, Kahtava noted.
Spectrum fragmentation compounds the problem. The outcome of WRC-27 will determine whether satellite transmissions on terrestrial bands get formal approval, and that decision is still roughly a year away. Without clarity on shared spectrum rules, some of the integration work remains speculative.
Masouros pushed back against the idea of a single unified architecture, arguing that the ecosystem is already heterogeneous and will only become more so as UAVs and low-altitude economy applications enter the non-terrestrial picture. “There’s not going to be an ideal, single architecture,” he said. “It’s going to be integration.” He pointed to AI-driven orchestration as one way to manage the complexity efficiently. Roberts, from the defense side, made a similar point: the realistic goal is common interfaces and common orchestration, not a monolithic design. Physical implementations vary too much across satellite links, cellular networks, and maritime communications for a single architecture to cover everything.
Kahtava offered a pointed lesson from previous generations. Both 4G and 5G ended up with too many options and too many bolt-on additions to the 3GPP standards. “We must avoid this non-standalone hassle,” he said, arguing that 6G should launch as standalone from day one. Getting a unified, commercially sensible solution that can be deployed across mobile broadband, IoT, and various verticals — without the fragmentation that plagued LTE — may matter more than any single technical feature.
Sub-terahertz, sensing and the business case
Jornet challenged the common assumption that sub-terahertz frequencies are simply about more bandwidth. When millimeter wave emerged for 5G, the pitch was capacity. Sub-terahertz for 6G risks the same reductive framing, and Jornet argued the real value lies elsewhere — in antenna gain. Smaller wavelengths mean that for a fixed antenna footprint, like the back of a smartphone, you can achieve significantly more gain. “Suddenly it’s easier to close a LEO link at sub-terahertz frequencies at 240 gigahertz than it is at 2.4 gigahertz,” he said.
The deep-space networking community made the same physics-driven move when it shifted to optical links — higher frequencies, smaller wavelengths, more gain over distance. But the satellite communications industry is already course-correcting on optics, Jornet noted. Optical links demand extreme pointing accuracy, which becomes impractical for anything that isn’t perfectly static. And contrary to the popular notion that terahertz signals can’t propagate through the atmosphere, the sub-terahertz range is more forgiving than its reputation suggests. Between 100 and 300 GHz, there are only two atmospheric absorption lines — at 118 GHz and 183 GHz. Systems operating at 225 GHz, for example, pass through the atmosphere better than optical links. Jornet pointed to FCC filings from major space companies as evidence that the industry is already moving in this direction.
Integrated sensing and communications drew interest as a practical enabler of the ubiquity vision. Masouros outlined how sensing could assist non-terrestrial infrastructure directly — tracking transmitters and receivers to enable faster, sensing-informed handovers that maintain connectivity during transitions between networks. He also raised the integration economics. Separate satellites for communication and remote sensing are the norm; a single satellite handling both functions would represent a meaningful cost reduction. On the defense side, commercial ISAC systems would operate under stricter power limits than dedicated military radars, but Masouros argued that geometric gains from distributed sensing across a connected infrastructure could recover some of those losses.
The harder question — who pays for all of this — didn’t get a clean answer. Kahtava was candid about the challenge. Satellite-terrestrial convergence offers limited commercial appeal in populated areas where terrestrial networks already work, and the business case for coverage in remote areas is unclear. “We don’t really know what is going to be the applications in five years from now,” he said. “So if we now plan for something, we must be totally off target.” Jornet offered a counterpoint on cost, noting that student-built satellites now cost under $50,000 and that modern LEO satellites are designed for three-to-five-year lifespans rather than decades. The economics of satellite infrastructure have shifted fundamentally, even if the business model for ubiquitous coverage hasn’t caught up. No clear killer application has emerged for 6G, and whether users in remote areas will pay enough to justify the investment remains an open question.
Table of Contents
