It seems that the idea of LPWANs (low power wide area networks) in unlicensed spectrum consuming the low-hanging IoT fruit, before the cellular providers could catch up, has failed to materialize. But while new LTE-M standards are poised for significant growth, there is one reason not to lose faith in an unlicensed alternative – the cable industry.
At Mobile World Congress, LPWAN specialist Sigfox posted encouraging growth to 13-14m subscribers as of the end of 2016, but during last year Jasper (now owned by Cisco) grew the cellular M2M subscriber base that it manages from 17m to 40m, and it expects its growth rate to be even faster in 2017.
As the 3GPP argued about a low power, M2M-optimized implementation of LTE, it seemed that the alternatives in unlicensed spectrum would be able to storm the market, carving out niches and securing big contracts.
Despite some successes, that has not proved to be the case and while there will be a valuable role for unlicensed technologies – for providers without spectrum or MVNO deals – there are doubts whether Sigfox, LoRa and the others have secured sufficient critical mass to enjoy significant growth outside of small niches. Meanwhile, the cellular operators have clung to existing M2M business using GSM, and now have their LTE-M and NB-IoT technologies to address more demanding applications which require 4G.
While these 3GPP Release 13 specs were a few months late, the market appears to have held out for them. Some MNOs have experimented with inhouse LoRa networks, and Telefonica even placed an order with Sigfox, but as those Jasper numbers show, it seems that many customers are happy to begin IoT roll-outs with SIM-based options, at Cat 4 and Cat 1 speeds, or to stick with GSM, before moving towards LTE-M later.
The two LTE-M options (Cat-M1 and NB-IoT) are on track to drive further growth in the cellular ecosystem, for management services like Jasper, Aeris and Kore, as well as the MNOs and their own IoT connectivity platforms and offerings. However, there are still issues facing these companies. It remains to be seen whether the software upgrade of existing LTE networks to support LTE-M will be as simple as hoped, and how quickly chips reach critical mass.
And many IoT applications require pervasive coverage, which does not exist for LTE in most markets, certainly outside urban areas. Whether IoT revenues ramp up sufficiently quickly to encourage MNOs to expand their LTE footprint into more rural areas, and into hard-to-reach locations like basements (important for smart meters), is uncertain.
The unlicensed spectrum alternatives may have the chance to address some services which are being held back by lack of LTE coverage – or the incentive to put that right quickly. However, the sector remains fragmented and it is hard to spot the kind of volumes which would significantly improve the economic proposition and give confidence in the long term viability of any one technology.
Part of this is down to the start-up nature of most of the unlicensed players. Sigfox provides the most definite numbers, and has grown from around 8m subscribers on the networks it powers in 2016 to 13-14m today. Sigfox told us that it was above 10m before its Telefonica deal, and that Telefonica had prepaid for a “few million” subscriptions. That volume could certainly ramp-up quickly if Telefonica’s IoT customers take a shine to the Sigfox-enabled Telefonica services.
Ingenu, a company with a very similar business model to Sigfox, is probably around the 4m subscriber mark, but it keeps its cards close to its chest. It is rapidly building out its global coverage, through the same kind of local network partner deals as Sigfox, but it isn’t disclosing the number of devices or its revenues.
Because LoRa isn’t pushed by a single company, it’s even trickier to estimate the total number of devices in the field. MNOs like Orange, Bouygues, and SK Telecom are building out national networks, but don’t report on the number of devices they serve. A large standalone LoRa-based operator, such as Actility, is in the same boat as Ingenu and Sigfox, a start-up which is not disclosing the number of subscriptions.
Consequently, a clear picture of the success of the unlicensed LPWAN vendors isn’t easy to come by, as disclosing that information would give outsiders a lot of insight into their books and valuation.
Nonetheless, there is one reason why licence-exempt LPWANs – or at least one of the candidate platforms – will survive, even if the M2M world is dominated by LTE-M. This is the rising need for unlicensed spectrum systems to support non-MNO providers, which will be important in a fragmented IoT in which private, vertically specialized networks are making a major comeback.
Nokia has recognized this trend and is heavily focused on supporting private networks with a cloud-based core network and with managed services. Its Wing and Impact platforms, geared to the industrial IoT and to vertical industry networks, support cellular and unlicensed M2M technologies, as well as WiFi, and can enable services run by MNOs, but also by other service providers, or directly by Nokia itself.
Given the huge variety of M2M applications which will need to be supported a decade from now, even MNOs are likely to want to use a combination of spectrum and technologies, in order to improve their overall capacity and costs, as they have in broadband with WiFi. Of course, LTE-M and future 5G may move into unlicensed bands as LTE-LAA has, but this does not yet appear to be a 3GPP priority.
But the big hopes for the LPWANs must lie in the vertical markets. Some companies, like Telensa, already operate mainly by providing private networks for cities or industrial companies. And some non-spectrum holders, notably cablecos, are eyeing ways to leverage their wireline networks, plus unlicensed spectrum, to get into lucrative vertical and IoT spaces.
At MWC, George Kakatsakis, VP of CableLabs’ advanced technology group, confirmed that the US-based cable R&D body has built a small LoRa network in order to examine whether this could support a new business opportunity for cablecos. The network, in the Colorado cities of Boulder and Denver, includes base stations and back end systems.
As a cable industry consortium, CableLabs essentially gets to decide industry standards for the cable TV market. The non-profit is probably best known for the DOCSIS spec, which provides broadband over cable lines, but also signs off on items like the WiFi standard for cable CPE. That remit means that CableLabs could end up defining the LoRa requirements for its members, writing the spec that would incorporate LoRa connections into CPE.
CableLabs has been running its LoRa network for about a year, determining whether it makes sense for cablecos to launch their own LoRa networks. LoRa could be used as part of a smart home play, though it would not be needed to connect the set-tops themselves, since DOCSIS does that. A greater opportunity would lie in overlaying LoRa on the cable footprint, by turning set-tops and home gateways into ad hoc LoRa gateways, and then using the resulting system as a readymade M2M network for vertical market and city services.
Like WiFi homespots, this could see cablecos deploying an inside-out network at very low cost by leveraging existing wirelines and customer premises. This might not be as optimized as a tower-based LoRa build, but it would be affordable, ubiquitous and adequate for many types of service.
The LoRa footprint would allow the operator to push smart home services, as part of a quad-play or quint-play offering, bundled on top of the broadband and TV offering. Once adopted, such an offering would make the customer much less likely to churn, as they would have to return all of the leased smart home devices that were provided as part of their monthly package.
The cableco might also be interested in selling access to the LoRa network to other businesses. Applications like home metering, asset tracking or smart parking could be served by a cableco with sufficient coverage, pitched at business, enterprise, or government customers looking for a low-cost network for a region-specific application.
In such a model, the cableco would be competing with the MNOs, but while those MNOs are currently better suited to national projects, a cableco-LoRa network would likely remain very competitive on price – as long as its target customers weren’t looking for a network that could roam over multiple states.
The balancing issue will be the cost of overhauling CPE designs to incorporate the LoRa gateways. Even if the LoRa functionality was provided by a peripheral device, a bit like the early WiFi bridges that linked set tops to home networks, it seems reasonable to assume that a cableco could cover a town with enough LoRa bandwidth for less than the cost of arranging tower deals and installations. Again, this would be a quick and dirty deployment, and far from optimum performance, but it would act as a gateway to diversifying into potential lucrative business agreements.
Back in October, Comcast announced an agreement with Semtech, the company that owns the LoRa PHY IP, launching its machineQ venture to explore the possibilities of LoRa. Comcast said it was launching trial networks in Philadelphia and San Francisco, with wider deployments over the course of the next couple of years, depending on the success of the trial.
machineQ will initially focus on utility metering, environmental monitoring, and asset tracking. Comcast is slowly coming to terms with the fact that it is now more of an ISP than a cable provider, and is looking to leverage its rather large fiver networks in the US – as backhaul for IoT services, as well as things like WiFi services to mobile customers.
In such deployments, the cablecos would be competing with the MNOs, who are in the very early stages of deploying the LTE rivals to the unlicensed LPWAN protocols – called LTE Cat-M1, and NB-IoT, under the umbrella-term LTE-M. The cable companies would also be going up against the likes of Sigfox and Ingenu, which are deploying their own LPWAN networks at a global scale, as well as from dedicated LoRa network operators like Senet and Actility, and MNOs that are turning their attention to LoRa, like Orange, Swisscom, and SK Telecom.
Last week, we spoke to TrackNet, a company that has just launched Tabs, a child safety and home security platform that uses a LoRa gateway in the home to monitor GPS-wearables and sensors. CEO Hardy Schmidbauer described the approach as inside-out, as TrackNet is looking to build home coverage first to provide wider area coverage, without having to rent space on a tower or install your own RAN infrastructure.
The deployment model for Tabs means that homes and day care centers (or any other building, by way of a WiFi-connected LoRa gateway) would be able to provide location and security alerts on a low cost network – cheap enough that TrackNet is able to sell Tabs in an upfront-only model, avoiding the monthly costs that Schmidbauer says put off consumers.
This inside-out model is similar to the potential cableco build-out, and TrackNet is having integration discussions with Senet, a North American LoRa network operator, and Comcast, to give its Tabs a wider network reach. In the EU, TrackNet is looking at operator partnerships too, and noted that European operators are eager for smart home devices like Taps to sell in their retail stores.
Schmidbauer says that the inside-out approach can reduce the need for towers by 70% to 80%, which saves a huge amount of capex for a startup such as TrackNet. For cablecos looking to use a similar model, they might be able to avoid towers entirely, if they can get enough gateways into their customer homes.
The CEO noted that the main challenge for LoRa roaming isn’t the technology, rather it’s the business model – working out how to share revenues between networks, or charge on a per-use basis in a way that makes business sense.