Home5GBehind the green curtain – Semtech reveals magic to merge cellular and non-cellular IoT

Behind the green curtain – Semtech reveals magic to merge cellular and non-cellular IoT

Note, this article is continued from a previous entry, available here

So what happens behind the ‘green curtain’, exactly? What is the trick that non-cellular IoT leader Semtech is looking to pull off with its purchase of cellular IoT maker Sierra Wireless? Or rather, how will the magic really work? Because the trick itself is known, already – traditionally as a single-tech bluff to manage two-way sensor comms in the cloud, and now, on paper, with the combination of LoRaWAN and LTE-M, as an expanded hybrid version of the same. But how

In conversation with Enterprise IoT Insights, Alistair Fulton, senior vice president of Semtech’s wireless and sensing products group (responsible for the LoRa radio technology and LoRaWAN wireless protocol), says, as before, the deal is less about joining cellular and non-cellular in hardware, and more about making them work the same way for developers. “It is about the horizontal platform layer, more so than the device – just because of the physics,” he says. 

He was asked on a conference call with analysts, after the deal had been confirmed, whether Semtech will look to build a single hybrid chip with a twin radio to run cellular LTE-M in licensed spectrum, alongside non-cellular LoRaWAN in unlicensed spectrum, he recalls. “And, you know, if I said that to one of our radio engineers, they’d fall over laughing because it’s not physically possible. It’s less about driving integration, and more about enabling choice.”

Which explains the trick, if not exactly how it works. But, quite helpfully, Fulton rewinds, as well, and says to look at how far the IoT industry has already come to simplify its ragged mechanics for the developer crowd – to make IoT easier to sell, install, and use. “If you go back 20 years, and consider all the challenges of IoT… well, kind of everything was difficult,” he says. The biggest headache, he explains, was the absence of easily available public cloud services.

“It was a nightmare,” he says. The game was mostly about data management – how to connect sensors, and gather and process data, just create order out of chaos. “Things have improved dramatically since then. On the cloud side, there are really powerful toolkits. I would hazard a guess the development time for a standard IoT solution in the cloud has reduced by 90 percent. Data is more manageable now. But connectivity is still hard.”

He explains: “Twenty years ago, everyone talked about 50 billion IoT devices (a reference to the infamous 2020 forecast by Ericsson and Cisco, actually from a decade ago, which, in ways, set the IoT market up to fail) – and we all meant cellular. Cellular was supposed to be the sledgehammer to crack the IoT nut. Which obviously wasn’t the case. Just basic availability of connectivity was a major challenge. And LoRa and other tech stepped in to fill the gap. 

Fulton – the challenge of IoT data management has been largely resolved; the headache of IoT connectivity management is still there

“But it has remained the case, also, that integrating these technologies is complex.” Repeatedly, he uses the example of a factory, typically sprung with a private IoT network, to describe this mess of application requirements and connectivity solutions – which is a different scenario, arguably, from the simpler tracking cases that LoRaWAN and LTE-M solve, separately, with public infrastructure. But the point is clear, and he goes on.

“There is all kinds of sensor tech in a factory – to track goods and machinery, to know if a conveyor has broken down, to know if a worker has fallen over, to stop a machine if someone’s arm gets caught. And all of those challenges require different radio capabilities. In an ideal world, a developer would not have to worry about what’s at the end point, and just write business rules that say: ‘If the data’s in this form, it needs to get to me at this time’. But that’s not the case.”

That is not the case, still, 20 years later. “Developers are still wrangling with all of these technologies,” says Fulton. Each side has strung up a green curtain, he implies; he references the MVNO magic-work by the likes of KORE Wireless and 1NCE to make cellular IoT a ‘no-brainer’ – global, affordable, manageable – for cellular IoT developers. Sierra Wireless is ahead of even them, he says (“at the forefront”), and Semtech has done the same with LoRa.

But these specialist low-power IoT operatives are working different venues, even as they are playing to the same crowd. “While we can make our own piece easier, developers are putting together a complicated jigsaw. Everyone likes to talk about single-tech use cases, but there aren’t any single-tech use cases, really. At the very minimum, you’ve got at least one radio – plus cloud and data systems, and all sorts. So there’s always this jigsaw. 

“If you need to use cellular with something else, it has been really quite difficult,” he says. Which makes it sound like the tie-up with Sierra Wireless will open a new channel for LoRa/LoRaWAN; which is right, but it goes the other way, too, of course. “We can and will do everything to make LoRa an easier tool to use, but unless we make that whole jigsaw easier to compose, then we’re not really doing our job.” 

Again, this is the magic, described with magnanimity, as if the only business sense for these old enemies to progress in the scrappy IoT game is to worry less about each other, and more about the market at large. As he said already, “It’s less about driving integration [and] more about enabling choice.” Indeed, the new logic for these factions, each failing separately to stitch-up a full-cover IoT fabric, is to knit together, make friends, and let the market decide.

This is what the Sigfox side is talking about of course, as it is reinvented under new ownership by Singapore-based Unabiz, currently touring the conference circuit on a ‘unified LPWAN’ ticket. Whether or not Semtech’s move on Sierra Wireless, to join LoRaWAN and LTE-M, plus bits of NB-IoT, further isolates Sigfox is another story, for another day. For Fulton, for now, it means some choice, at least, between licensed and unlicensed IoT. 

Integration? Maybe, sometimes, he says; but choice is the thing. “In some ways IoT has suffered; differentiation is not a good word in IoT. Developers look at it and go, ‘Oh my God; that means this doesn’t work with that’. And normalizing that differentiation is tricky. Other platforms have tended towards the lowest common denominator, and lost some capability from the tech. Our goal is to keep the special aspects of each, and make them available via a simpler platform.”

Back to the green curtain, again. But just on hardware integration first; Fulton notes that, despite the physics of twinning IoT radios at chip level, the market will engineer hybrid IoT into dual-mode modules anyway, which switch between network technologies on-the-fly, as required. “There will be solutions that do that. We see them today, actually; there are customers out there today building trackers that combine LoRa with cellular,” he says. 

It is not that hardware integration is not workable, or needed. But the demand is fitful, the engineering is tough, and there is an easier way. “We do see these hybrid devices that pick and choose, and developers walk over broken glass to build them. But in the main, honestly, it’s about being able to build one solution. It’s back to that factory example – one solution that comprises all different sensor types, and sets whatever data rules are needed.”

Which means, effectively, a design-once IoT solution that can be orchestrated many ways, on any (different) connectivity networks, whether to build a picture of industrial wear-and-tear for periodic maintenance, or to halt an industrial machine in a heartbeat to save a life or rescue a workload. “That is where we see the real value,” says Fulton. There’s the pledge, and there’s prestige, in magic-speak; but, again, what about the turn?

What has to happen so that developers can build-once for cellular-based LTE-M (and NB-IoT, here and there, perhaps) and non-cellular LoRaWAN? Who holds sway? Will Semtech impose the LoRa way on the Sierra-designed cellular market, or will it take from the Sierra Wireless model for IoT provisioning and management in the cloud? How does this work? Provisionally, what is the process from now, to integrate the companies and align the technologies?

Fulton moves as if to pull back the curtain. He responds: “If we end up implying major change in either of those ecosystems, then we will have failed. Developers want to do the same sorts of things, even if they happen in different ways. The way to update firmware, say, is very different for each – because of compute power and memory, and bandwidth. Those differences will remain because they are inherent to each technology. But the goal is the same.

“Making the process simpler for developers, while enabling them to leverage differences in the tech, is the north star here. That is where the work begins. And on the licensed side, Sierra has the leading platform to get the job done, across a range of different tools – from MVNO connectivity provisioning to data management. On the unlicensed side, we have been focused on device provisioning and management, and more recently device positioning.”

He goes on: “We’ve got pieces of the jigsaw and they’ve got pieces of the jigsaw, and unusually – and I’ve done a lot of tech acquisitions over the years – there is no overlap. These capabilities are complimentary. There is work to be done to bring them together in a common platform. It’s more than just writing a fancy UI. But I have as much confidence, as is possible, in bringing together these two highly-seasoned, highly-skilled engineering teams.”

But the minutiae of corporate integration is not the thing to ask about, says Fulton; it sounds like a cliché, but the developer ecosystem (“the customer”), at both shows, has to direct the process. As the background to this tale tells, toxic in-fighting and vain land-grabs have seen the tech-side wag the IoT dog, to now. What happens behind the curtain should be designed with solution makers; they are the ones in the tall hats and long coats, after all.

Semtech and Sierra Wireless are only making the props. Fulton responds: “The bigger answer is to focus on what the customer is trying to do. The first step is to work with partners to say, ‘We think this is the objective, and this is how it should work – but tell us where we are wrong, and what you think’. The second step is to look at the portfolios and capabilities of both companies, to figure out how to deliver that goal. What is the service combination?

“And the third step is to decide where we have it covered and where we have a gap – where we can just stick it together, and where we need to build a bridge. And that process is going to take a bit of time. We have been engaged with the Sierra team for a good while, but it is only now that we start to really get into the nitty gritty about how to pull this together. But with the teams in place, I have no concern whatsoever about our ability to do that.”

Does that explain the turn, which vanishes the complexity of hybrid IoT behind a green curtain? Maybe not; but only because the rehearsal has just started – and it explains the process for that, and the concept of the trick, itself. A couple of other questions to consider, though. Firstly, it seems like there is a potentially delicate balancing act for Semtech with the acquisition of Sierra Wireless with its own stable of go-to gateway manufacturers? 

How will that work? What are the likely terms? Will there be clearly defined roles and responsibilities, where Sierra Wireless, as the in-house gateway maker, sticks with cellular hardware, or offers something different in a hybrid form? How will the old guard, which has grown LoRa for Semtech, and paid royalties in the process, know Sierra Wireless is not the preferred Semtech production house? What is the message to them?

“Well, I don’t think it is an unusual situation, for a start. I mean, it is something Sierra has today, already. One of its biggest customers is Cradlepoint, which produces its own routers – which compete with Sierra’s routers. And that relationship has been managed very well over the years. At the end of the day, choice is what is important. Yes, there are opportunities where the addition of LoRa to a cellular gateway makes sense,” responds Fulton. 

“There are some use cases where, at the hardware layer, a hybrid device offers value; less so at the end point, just because devices tend to be one thing or the other – for cost and power reasons. But we don’t have any intention to compete with our module customers. It is about bringing this together in a common cloud layer. Some of the commentary has been about moving up the value chain to deliver [hardware] solutions. Which is not the case.

“This is about providing a horizontal platform to make it possible for developers to use whatever tool they want – whether that is cellular LPWAN, or standard 2G/3G/4G/5G, or LoRaWAN. To be clear, the reason we are doing this and the thing we bring to this party – versus someone else acquiring Sierra – is LoRa. We see this as a way to democratize access to LoRa, and to drive very significant growth in the LoRaWAN ecosystem. 

“Sierra has a formidable range of customers. Putting LoRa into that toolkit means they can take advantage of LoRa, where cellular falls short. That is the goal – to take away this island choice. Because right now, they have to decide where to hitch up their boats, and the distance between these islands is significant – in cost and complexity. We want one unified LPWAN island so developers are not constrained by the macro level choices they’ve made.”

The other question is about the role of NB-IoT as part of a future-joined business? Because NB-IoT has always been in the targets of the LoRaWAN brigade; where LTE-M, its cellular IoT twin, has been traditionally pitched for for higher-fidelity IoT use cases, lower-power NB-IoT has always been painted into a tighter corner with LoRaWAN and Sigfox, and the like. Is there, by chance, any kind of a plan to kill it, at least in the Sierra Wireless portfolio?

Fulton appears magnanimous, again, making the same point about letting the market make up its own mind; but there is more to his answer than just that, maybe. He responds: “It is not our job to decide what customers can and can’t use. There are scenarios where, in the future, if NB-IoT is deployed much beyond China, then it will become a more valuable tool for developers. And it is our responsibility to make that possible. 

“But to be very clear, LoRaWAN is a better tool than NB-IoT in most situations. And we think people will choose LoRaWAN more often than not. At the same time, I’m not going to say, right now, that we won’t continue the work Sierra has been doing with NB-IoT. We may have a different view in a few months, but, right now, our goal is to maximize choice and ease-of-use. Developers will pick the right tool, and we think they’ll pick LoRaWAN.”

Previous post
Vodafone Idea seeks partners to boost enterprise segment: Report
intelligent transportation systems
Next post
Euro transport and space agencies team up on satellite comms for smart vehicles