NTN’s unsolved problem: proving seamless connectivity on the device

NTN roaming, spectrum and 6G gaps all land on the handset

In sum – what we know:

• Policy-driven handoff – Verizon said seamless NTN means attaching to the right network on cost, capacity and throughput rather than just holding any connection, with enterprise customers demanding more.
• Standards gaps remain – 3GPP Releases 17 through 20 lock in the NTN architecture for 6G, but roaming and load-sharing between multiple satellite providers still has no standard and depends on non-standard integrations.
• Interoperability is the bottleneck – Rohde & Schwarz expects devices not to work out of the box across networks early on, and better cost per bit will require more spectrum shared intelligently between terrestrial and non-terrestrial operators.

Seamless connectivity is the promise on every NTN marketing deck, and RCR Tech’s NTN Integration Forum, held September 8, spent a full session taking the word apart. Initial acquisition and attachment are largely solved problems. What is not solved is quality-of-service assurance as a session moves between a mature terrestrial network and a satellite network whose cells are moving at orbital velocity — and, beyond that, mobility between one satellite provider and another, cross-border roaming deeper than anything terrestrial roaming offers today, and the intelligence to choose the right network for the right session without exposing any of that complexity to the customer.

Much of that complexity lands on the device. The user equipment has to locate satellites and pre-compensate for their delay and Doppler using GNSS and broadcast orbital data, carry filters for bands most phones do not have today, and interoperate across satellite and terrestrial networks whose interworking is still unproven.

Overcoming NR-NTN integration hurdles to deliver seamless connectivity

RCRTech principal analyst Sean Kinney opened the panel by noting that integration between terrestrial and non-terrestrial networks is usually discussed in the abstract, when device attachment, session continuity, authentication, policy and roaming are all layered inside it: “it’s not as simple as it may sound on a marketing deck.” He asked XRComm founder and CEO Dr. Ahsan Aziz; Kevin Cohen, VP of direct-to-device business development at Viasat, who also brought the perspective of the Mobile Satellite Services Association; Obilor Nwamadi, senior product manager at VIAVI Solutions; and Jyoti Sharma, associate director of network planning at Verizon, which piece is hardest today.

Aziz set the benchmark at the experience users already take for granted when they walk from a cellular network onto Wi-Fi. Device connectivity and initial acquisition are largely solved, and much of the rest is governance and policy. “But the quality of service assurance as we go from one network to another network, a highly mobile network, if you will,” he said, “that still remains to be a big challenge and a problem to be solved.” Sharma reframed seamlessness from the operator’s side: “It’s not just constant connection, but it’s really policy-driven attachment.” Verizon’s task is to hand customers off to a satellite network and bring them back to terrestrial when conditions are right, “based on cost, based on capacity, based on throughput, based on their KPIs,” and enterprise customers “have even more stringent requirements” than consumers.

Cohen located the prize somewhere smaller than the coverage map suggests. “Some people think of, hey, I don’t have cell service, I’m going to be connected over satellite. But really, the long-term goal is going into those small patches where there’s no service” — the 30-second white spot on a drive that drops a voice call. Because capacity from any one non-terrestrial network varies by cell and location, “one of the requirements is going to be that there’s multiple non-terrestrial network providers that are offering service over those cells where there’s lots of demand,” which is why MSSA’s goal extends to mobility “from non-terrestrial to non-terrestrial.” The standards are not there yet. “If you look at the 3GPP standards, they aren’t all the way there. So either you have to do a non-standard integration with a cellular operator to get that seamless mobility, or you’re going to have a connection that still breaks.” Nwamadi added the device dimension: receiving a signal from GEO is different from very low Earth orbit and the bands differ, so customer equipment ends up being designed for FR1, FR3 or FR2 depending on the service — a problem that “is not fully cracked yet,” down to how a terrestrial antenna and an electronically steered antenna coexist in one form factor.

On the gap between standards and field reality, 3GPP walked through the release map: Release 17 introduced the general NTN architecture, Release 18 concentrated on interconnectivity and mobility, Release 19 opened the regenerative architecture alongside the transparent one, and Release 20 lays the groundwork for 6G. The reassurance for anyone investing now is continuity — “none of that will be lost in the transition to 6G,” since 3GPP has already said the Release 17 architecture and its waveforms are here to stay. What 6G adds is new dimensions, above all positioning, navigation and timing from communications satellites. Today a handset needs a GNSS fix to connect to an NTN satellite at all, because it must know its own location and the satellite’s to pre-compensate; Release 21 will try to unpick that, lifting a constraint that forces IoT devices to carry a GNSS receiver that drains the battery faster than the radio does — and, Cohen noted, leaves them unable to connect at all under the GNSS jamming seen in recent events. Nwamadi said VIAVI’s 3GPP delegates report little appetite for changing the waveform in 6G, “to be able to keep backwards compatibility with 5G,” especially as “NTN came mid-5G cycle, so it hasn’t even lived a whole cycle of a generation.”

Standards do not put features in phones, though. “You can have all the standards in the world, but the chipset providers have to decide to put those into devices,” Cohen said, and they do so when an operator buying millions of devices demands it. Even for the standard MSS L and S bands, “there’s only a few standard smartphones that actually have those bands in them today,” and even a very small cost delta to add them adds up at scale. Sharma agreed the ecosystem is the work. “For MNOs like Verizon, this is not new,” she said; Verizon has been developing networks based on 3GPP standards for decades. What is new is that “there are two big ecosystems that are coming together,” from chipsets to core to satellite network, “and one player is not going to do it all.” Verizon is already testing and deploying narrowband IoT NTN, with devices being certified.

Aziz, whose company’s stated mission is closing the gap from lab to launch, explained why that gap persists. A standard tells you how things should operate; interoperability is proven through IODTs, plugfests and certification, and because the time any party gets with a satellite is limited to its visibility window, chipset, network, satellite and operator teams end up in silos. Common dashboards and tools available across the industry would help all the players align, and “that goes a long way.” The same logic governs ground truth. Simulation defines the scope and conformance defines what a device may do, “but the actual field performance is something different,” and there is no clearly defined spec for it. As architectural splits multiply, “the scope for testing or the simulation scope explodes and becomes intractable,” and while digital twins help, a twin is only as good as the data it gets from the ground. As Aziz put it, citing a famous saying, “all models are wrong, and some of them are useful.” Nwamadi described how twin creation “is being slowly decoupled from that need for hard ground truth data” — though ground truth is “still useful for calibration” — through ray tracing borrowed from the gaming industry, mixing site-specific twins baked into RF test equipment with pure-software twins at city scale, “where the constraint becomes compute and not necessarily RF connectivity.”

How does an operator choose among all this without passing the complexity to the customer? Sharma said Verizon wants to take that complexity out of its customers’ hands and “make sure that the network can make intelligent decisions” — GEO, LEO or MEO chosen at the network layer against the customer’s requirements and exposed through APIs. “That is not currently defined in the standards. That is really innovative work that the operators need to do.” Cohen’s answer to an audience question on roaming showed how far there is to go: “I don’t think that, especially in Europe, that terrestrial connectivity works well for cross-border connectivity,” and the multi-provider NTN case will need “a much more deep integration, which is not how integration goes today” with IPX providers and standard roaming agreements.

Kinney closed by noting that one multinational operator has six satellite relationships; the goal of convergence is to avoid a terrestrial network with siloed satellite services alongside it. Cohen’s answer was standards that do not exist today for roaming between NTN providers and for distributing load across networks as needed, by “cost basis, capacity, capability, even actual traffic that you’re tagging” — an IPX-style layer for NTN that MSSA intends to help build. Aziz argued that real convergence will come when the terrestrial network “looks at the satellite network as a neighbor cell and vice versa,” with PNT as “a first-class citizen.” Nwamadi drew the line most bluntly: “I personally think the technical parts, not a lot of things left to be solved. I think it’s mostly solved.” The work now, he said, is on the commercial angles. Sharma had the last word and the clearest demand signal. “We hear from our customers no connectivity is not an option, and we hear it loud and clear. Every 5G request comes with NTN.” Delivering it takes the whole ecosystem working toward a 6G in which NTN is native rather than “an afterthought like we did now in 5G.”

Don’t forget the user equipment: a test and measurement view

Picking up the UE thread the panel had seeded, Colm Mulligan, market segment manager at Rohde & Schwarz, organized the landscape into four flavors of NTN. On one side sit non-3GPP systems, “a mixture of open and different proprietary standards.” Next is direct-to-cell — what the FCC calls supplemental coverage from space — which “enables different legacy devices who know nothing about talking to satellites to have satellite connectivity.” Then come the IoT NTN standards, most prominently narrowband, and finally NR-NTN, “which has in the future the promise of higher data rates.”

The market dynamics differ across those flavors. In direct-to-cell, Starlink has launched with a number of operators and AST SpaceMobile is coming, and Mulligan found the most interesting aspect to be “the way these two different companies from terrestrial and non-terrestrial sides are working cleverly together to share spectrum in specific markets.” Deutsche Telekom has announced launches with Starlink in eight to 10 European markets from 2028. A wave of mergers and acquisitions across the sector — Rocket Lab, Iridium, Amazon, Globalstar, Lynk and Omnispace, Space42, Viasat, SpaceX and EchoStar among the names involved — has a common motive: “one of the key things driving all this is spectrum and the desire to get access to valuable spectrum.” In Europe, IRIS² is advancing on an accelerated schedule as the region works through how it will regulate the MSS band.

Spectrum also sets the economics of the early market. The lower 3GPP bands in L and S band offer “quite small bandwidths — we’re looking at 10, 20, 30 megahertz of bandwidth for the channels.” Ka and Ku bands from 10 to 30 GHz offer the future outlook for capacity, but initial NTN deployments will concentrate in the lower frequencies, and because of those constraints “the cost per bit will be somewhat higher” than users are accustomed to on terrestrial networks. The European Commission has launched a funding program to incentivize development of Ka-band user terminals for 3GPP standards. Mulligan’s timeline placed non-3GPP Ka/Ku services from Starlink, Eutelsat and Amazon Leo as live or imminent; direct-to-cell live with Starlink and expected from AST early next year; and IoT NTN live with Skylo, Sateliot and OQ Technology, with Iridium expected later this year. “NR-NTN services, we would expect to see launching in late 2027, early 2028 in different LEO constellations, and the first Ka/Ku launch is still a number of years away yet for NR-NTN.”

On the technology, Mulligan concentrated on the two physical effects that dominate device behavior. Satellites range from high-altitude platforms at about 20 kilometers to geostationary orbit at 35,000 kilometers, and that spread drives round-trip delay. A satellite appearing on the horizon presents an initial delay that defines the two-way latency of the connection; it shrinks to a minimum as the satellite passes overhead and grows again as it exits. “This longer and changing delay affects many different aspects of the modem performance in the device and the application, and it needs significant testing in a controlled environment.” Doppler follows the same arc, changing continuously as the satellite moves and most extreme in LEO. The test portfolio he described tracks the device life cycle: antenna characterization for traditional and phased-array designs, modem development and integration including GNSS signal generation, interoperability and certification testing across satellite and terrestrial networks, and automation for production and field test. At the center is the CMX platform, which supports all three of the key 3GPP technologies in one multi-orbit, multi-band solution, from RF and protocol measurements to application testing, with NTN mobility procedures for inter-beam and inter-satellite handover and built-in fading and channel emulation.

Asked to distill the biggest challenge device manufacturers will collectively face, Mulligan pointed to interoperability. “To work across all these different networks from the device perspective is not a given.” He recalled how hard interoperability between networks was as a young engineer 25 years ago and expected history to repeat. Devices cannot be expected to work out of the box across all these networks in the initial phase, “and I think that’ll be maybe a surprise to some people.” On the path to a better cost per bit, he split the answer in two. “We will need more spectrum, but we will also need to intelligently share that spectrum between terrestrial and non-terrestrial operators. That will be key, and we’ll need to do that on a global scale.” The second part was personal: a passionate mountaineer, he hopes smartphones will eventually reach the higher frequencies so that he can be on a summit and still reach his work. “With a combination of innovation and regulation and intelligent sharing of spectrum, I think we go a long way to improving the cost per bit.” The device is where the forum’s abstractions about seamlessness are either proven or exposed — which is exactly what the operators deploying today have found.