GSA charts Cat-M and Cat-NB surge, LoRa’s public-private gulf narrows

The Global Mobile Suppliers Association (GSA) has announced that 102 MNOs in 52 countries have deployed one or both of the two Licensed LPWAN (L-LPWAN) technologies LTE Cat-M and Cat-NB. It comes as the LoRa ecosystem begins to shift from private LoRa networks to public ones that support the open LoRaWAN specification, with Semtech saying that the gulf between its private LoRa and LoRaWAN shipments is closing.

The report focuses on MNO adoption, rather than footprint coverage or the number of devices. To this end, 149 operators in 69 countries are investing in L-LPWAN, with 102 MNOs in 52 countries having actually pulled the trigger on a deployment. Of the 102 MNOs, just 20 MNOs across 19 countries had deployed both.

Cat-NB seems to have a distinct lead though. The GSA says 140 MNOs in 69 countries are investing in Cat-NB, compared to 60 MNOs in 35 countries investing in Cat-M. Similarly, 88 MNOs in 50 countries have launched Cat-NB networks, with 34 MNOs in 24 countries having launched Cat-M. This leaves 28 countries having only Cat-NB, with 2 countries home to just Cat-M – Mexico and Argentina.

So, 55 countries have Cat-NB coverage, to some degree, and 24 have Cat-M. Some 22 countries feature both a Cat-M and a Cat-NB network, with 20 MNOs having coverage for both in 19 of these 22. The LoRa Alliance, the initiative behind promoting LoRaWAN (the open parts that build on top of Semtech’s proprietary radio silicon), says it has 100 network operators live in 51 countries, with 100 countries total having LoRa deployments.

Both camps don’t report coverage footprints in detail, nor do they outline how many base stations and radio units are involved. To this end, a country might count as covered if just one base station was live, but counting that would be somewhat disingenuous. It would be good to see more detailed accounts of the infrastructure deployed, but that doesn’t seem likely at all, in the near future at least.

Sigfox meanwhile, to its credit, does have a map that you can drill down into, to determine if a particular area is covered or not – but again, no detail on network infrastructure deployed. That map will look a whole lot different if the Eutelsat satellite deployment comes to pass, which would drive Sigfox up from its current presence in 60 countries.

On the L-LPWAN side of things, there’s still a lot of work to be done in roaming agreements, so that devices can pass between territories and different operator networks without losing a connection. The bigger issue is installing network coverage in the first place, but once that is underway, the practicalities of tying all the different MNO’s back-end systems together needs to be completed. This is a must-have requirement for any international application, but of course, a much lower priority for a domestic or static deployment.

It seems that LoRa’s biggest future headache is how to ensure a device can roam between the different LoRa radio regions, or rather, the mechanism for detecting when a device has left Europe and arrived in the US, for instance. In theory, the packets sent out should make their way back to the right application, through the LoRaWAN cloud environment, but if the device is broadcasting on the wrong frequency, it will simply be chirping into the ether.

But getting all the LoRaWAN network servers (LNS) hosted in the cloud to talk to each other is perhaps another architectural challenge. With The Things Networks (TTN) being perhaps the most expansive public LoRaWAN network, there needs to be a clear and simple way for the TTN back-end to tie in with the likes of Actility and Everynet ecosystems. These concerns were outlined in the LoRa Alliance’s Backend Interfaces Specification, which was used by Orange, KPN, and Actility to demonstrate roaming between the Netherlands and France in February 2018. The next major step will be the aforementioned roaming between different RF regions.

As the LoRaWAN ecosystem works on that step, it will be paying close attention to the shipments of LoRa silicon from Semtech. In the early days, Semtech was shipping the vast majority of LoRa chips to customers that were setting up their own private networks. These customers may well have not been covered by a public LoRaWAN operator, and so opted to install their own network infrastructure. Security concerns are another motivating factor, as are per-device costs – as a very dense deployment could see rather off-putting per-device fees if that user has to rely on a network operator.

At MWC, we learned that the gulf between sales for private LoRa networks and LoRaWAN networks is closing. Of the 80mn LoRa chips shipped, around 80% used to be used in proprietary LoRa environments. Today, that figure is around 60%, and with the projected increase in demand for LoRa chips, it won’t be long before the majority of all shipped LoRa chips are active on LoRaWAN networks. It’s in Semtech’s interest to foster demand for LoRaWAN, as the semiconductor business is a volume play – its needs to sell as many chips as possible, and a more open ecosystem is conducive to this.

And so, it looks like the chief concern for the LPWAN market in the next couple of years is going to be solving the question of roaming – demonstrating to the massive global customers that the technology being slung is capable of supporting an international application. Of course, Sigfox’s in-house single-network system takes a different approach, having solved roaming from the get-go, but it also plays at its ‘0G’ level, with small overlaps with the rest of the LPWAN community.

Satellite incursion may be the most pressing issue for LPWAN in the near future, as a swathe of new IoT-focused entrants have drummed up funding and plan to launch nanosatellite constellations. These networks, like Sigfox, won’t have to worry about roaming, as they will be a single global operator, but could lose out on the economies of scale that can be enjoyed by the larger LPWAN ecosystems.