In search of the low-power wide area network standard for IoT

One of the major obstacles holding back the mass rollout of M2M/IoT applications has been the lack of an accepted standard for low-power wide area networks. But now there’s a number of options available with more in the works. James Atkinson looks at what’s on offer

Until recently, end users of machine-to-machine (M2M) and Internet of Things (IoT) applications have largely had to rely on either wired or licensed cellular connectivity solutions provided by mobile network operators (MNOs).

There is no question that 2G (which accounts for around 60% of all cellular connections), 3G and 4G are suitable for some IoT applications. But for many other applications they are too power hungry and much too expensive. A white paper from NextG-Com and Arm, entitled Redefining LTE for IoT, published in September 2014, cited the current cost of a 2G device module as about $10 and for an LTE module $40.

As a result, the search has been on to develop low-power, low-cost and generally low-throughput solutions suitable for IoT applications. Several candidates have emerged in the short range (in the home) sphere such as Wi-Fi, Bluetooth and Bluetooth Low Energy (BLE), ZigBee and Z-Wave among others. But most of these are extremely difficult to scale up to handle high densities of end points, and their short range means they are no use for wide area networks.

Definitions and goals vary for low-power wide area network (LPWAN) solutions, but there is general agreement that they require: very low cost devices (module cost of less than $5, others say it needs to be $2 or below); and a battery life of more than 10 years (with, say 2 x AA batteries, and assuming 100 bytes/hour uplink).

Other goals include: the ability to support large numbers of M2M devices; an additional 20 dB in range (in excess of 160 dB) compared with existing cellular technologies to ensure reliable coverage in hard-to-reach locations; and an application payload size of typically between 20 and 200 bytes.

A number of companies and industry bodies are working to meet these challenges. These can be divided into solutions developed by start-ups ranging from free, fully-open standards to proprietary or semi-proprietary ones, and then there’s a number of options being looked at by standards body 3GPP in the licensed cellular space – most of which are derivations of 4G LTE.

The independent options

If you want an LPWAN solution right now that meets the above criteria you are not going to find it on licensed spectrum. This has given the independents a first-mover advantage. The main contenders are: France’s Sigfox; the LoRa (Long Range) Alliance, spearheaded by California-based Semtech Corporation; Ingenu (until recently known as On-Ramp Wireless), also based in California; and the Weightless Special Interest Group (SIG).

In terms of deployments, Sigfox is probably the most established player. It uses ultra narrow band (UNB) technology in unlicensed spectrum (the 868MHz ISM band in Europe; 902MHz in the US) which means it can only transmit a very small amount of data per day due to power emission regulations in the unlicensed bands. It is largely a one-way uplink dominated communication network.

However, it has quite a number of customers in Europe, including Arqiva in the UK, which is using it for its Government smart meter contract. It launched in the US in March with ambitious plans to cover 90% of the country within three years, backed by a recent $115 million funding round.

On-Ramp Wireless, which changed its name to Ingenu at the CTIA Super Mobility show in Las Vegas in September, also announced plans for a national US rollout of its Machine Network. Ingenu uses a protocol called Random Phase Multiple Access (RPMA) operating in unlicensed 2.4GHz spectrum, which deploys self-modulating techniques to mitigate interference from other users.

The RPMA technology supplies a full two-way link, and the company describes its raison d’être as being about providing better coverage (it claims its towers can cover more than 200 sq miles), but not faster speeds when compared with existing cellular networks. Ingenu has raised more than $100 million in funding to date.

The LoRa Alliance was formed in January this year as an ‘open, non-profit association of members’. In June 2015, it released the LoRaWAN R1.0 specification. This uses a modulation format based on Chirp Spread Spectrum (CSS), which is widely used in radar and ranging applications, but not commonly deployed in communication systems because of its susceptibility to interference.

LoRa has developed three classes of end-point devices: bi-directional with two, short downlink receiver windows; bi-directional with scheduled receive slots; and bi-directional with maximal receive slots (nearly continuously open). LoRaWAN data rates range from 0.3kbps to 50kbps. To conserve battery life the LoRaWAN network server manages the data rate for each connected sensor via an adaptive data rate algorithm.

Another player operating under the LoRa Alliance standard is Link Labs, but it has also developed a proprietary system to provide more advanced functionality called Symphony Link. It uses the LoRa physical layer, but not the stnadard LoRaWAN MAC architecture. 

It claims a range more than 100 times that of Wi-Fi and at a cost well under that provided by cellular networks. The Symphony Link gateway is an 8-channel base station operating in the 915MHz ISM band in the US and 868MHz band in Europe. Modules will require LoRa chipsets and Symphony Link specific software.

Each of these protocols is, to a greater or lesser degree, proprietary. The Weightless SIG, on the other hand, was set up as a free, open standard from the start. It originally targeted white space spectrum in TV guard bands (particularly between 400-800MHz) with development spearheaded by UK start-up Neul.

Weightless-W uses a star network architecture with data rates ranging from 1kbps to 10mbps depending on link budget, with data packet sizes from 10 bytes and no upper limit. It provides full two-way links and offers a range of 5km in urban environments and up to 30km in line-of-sight rural locations.

However, with the exception of the US, regulators around the world have been very slow to open up white space bands, so Neul decided to switch its efforts to licensed bands and was then gobbled up by Huawei in September 2014.

The Weightless SIG then switched its attention to the ISM bands below 1GHz after UK firm NWave Technologies decided to make its IPR in that space available to everyone. Weightless-N has similar characteristics to Sigfox using unlicensed UNB technology. Weightless-N networks have been deployed in London and two Danish cities this year. A full and free SDK kit was launched by NWave in early October 2015 - see full story here.

Weightless-P is the SIG’s latest variant, which is being developed on the back of Taiwanese company M²COMM’s decision to open its IPR up. This is a full two-way link, ultra-high performance LPWAN solution using ISM/SRD bands with an adaptive data rate ranging from 200bps to 100kbps. Base stations, endpoints and development kits are expected to be commercially available in Q1 2016.

Most of these solutions have the advantage of being available right now, but they do have some disadvantages. Jamie Moss, principal analyst, consumer technology & Internet of Things at Ovum, observes: ‘It is very hard for small start-up vendors, especially proprietary ones, to gain acceptance because it requires a lot of faith to buy into a new ecosystem and believe that it will be there in the future.

‘I think that these kind of companies start up for the sake of being acquired by established players once their offering is fully formed, and they can then be integrated into the market via a well-known manufacturer, who everyone knows and trusts will be around for a long time. Neul was a good example of this, being snapped up by Huawei,’ says Moss.

‘The Weightless SIG approach of providing fully open standards is an interesting way to go to market by making it all free. I think they are hoping that the likes of Ericsson or Huawei might pick it up and run with it and give the market confidence in the standard and start to get orders,’ he suggests.

The LTE-MTC options

Turning to the cellular world, several alternatives are being looked at to try to adapt licensed 4G technology, in particular to meet the performance criteria set by the independents outlined above.

The first approach being worked on by 3GPP is to take the LTE standard and adapt it to meet the LPWAN aspirations outlined earlier – referred to as LTE-MTC (machine type communication). The starting point is LTE Release 8 Category 1 (Cat-1) where devices have a downlink peak rate of 10mbps (uplink 5mbps) and are full duplex.

Release 12, published in March this year, introduced a new lower complexity Cat-O device. This has a reduced peak rate of 1mbps for both downlink and uplink, achieved via reduced transport block size, half duplex capability and only one antenna.

3GPP estimates Rel. 12 can reduce the bill of material of a Cat-O LTE module by 50% compared with a single band Cat-1 device. Rel.12 also introduced Power Saving Mode (PSM), enabling devices to enter sleep mode, thereby saving power, for long periods of time. It also established a way to enable the 20 db coverage improvement.

Release13 (due out March 2016) will reduce the complexity still further by defining a Cat-M, which will offer uplink and downlink peak rates of around 200kbps, half duplex and with a modem 75% less complex and less costly than Cat-1 devices.

It will also be the first to provide a reduced device-receive bandwidth of 1.4MHz, compared with 20MHz for Categories 4, 1 and O – this is where the main cost reduction comes from. In other words, the LTE spectrum is transmitted in narrower slices, making better use of limited spectrum.

Further reductions to just 200kHz, narrow band in other words, are also being studied, although a 2014 white paper, LTE Evolution for Cellular IoT by Ericsson and Nokia Networks concluded: ‘Introduction of a narrower LTE system bandwidth (eg. 200kHz) can be considered but requires substantial additional efforts to the improvement listed above (ie. those envisaged by Rel.12 and 13).’

As regards timelines for commercial deployment, the major M2M module providers u-blox, Gemalto’s Cinterion and Telit – working with chipset provider Altair – have only just unveiled their first commercially available Cat-1 IoT modules this month (September 2015), while Sierra Wireless’ first Cat-1 module is due shortly. However, commercial deployment of Cat-0 is not expected until 2017 and Cat-M in 2018.

The Clean Slate alternatives

But not everyone is convinced that taking a standard designed for high power, high data rate transmission is the best approach to finding a cellular LPWAN. This has led to the so-called ‘Clean Slate’ solutions, which have split into two camps: NB-LTE (Narrow Band LTE); and NB-CIoT (Narrow Band Cellular Internet of Things).

Huawei and its acquisition Neul argue in a white paper entitled "Introduction to ‘Clean-Slate’ Cellular IoT radio access solution" that: ‘These benefits (ie. the LPWAN criteria set out above) are very hard to achieve through the evolution of existing cellular radio access technologies’ and that this is ‘because the IoT requirements are so different from mobile broadband’.

The Huawei/Neul approach to NB-CIoT is set out in the above white paper. In the other camp are Nokia, Ericsson and Intel, who announced on 14 September that they were plumping for NB-LTE. NB-LTE outperforms even LTE-M by providing downlink peak rates of 200kbps in just 200kHz of spectrum.

The three companies argue that by using their approach, MNOs can re-use much of their existing LTE network technology and chipsets, thereby facilitating a quicker route to market and taking advantage of economies of scale.

One final contender in the cellular LPWAN stakes is EC-GSM (Enhanced Coverage GSM), which takes the same approach to adapting 4G, but applies it to 2G GSM. This would enable MNOs with legacy 2G networks to provide a cost competitive LPWAN option using existing infrastructure.

Regarding the various cellular options, Daryl Schoolar, principal analyst, Intelligent Networks at Ovum, says: ‘I think there is a real possibility that Cat-0 might be skipped over for Cat-M. Cat-1 should be available end of this year. Cat-M availability should be in 2017-18, so why bother with Cat-0 in between?

‘The big obstacle I see with the clean slate option is that it requires new infrastructure investment, while what Ericsson, Nokia and Intel are proposing is a software update, which seems like a better prospect for mobile operators than buying more radios.’

Which standard?

So, what to make of this plethora of LPWAN options? Is it a case of survival of the fittest? In which case, what standards will win through? Or does the industry need a number of different standards? Opinion is divided.

Professor William Webb, CEO, Weightless SIG, observes: ‘There’s now a whole range of LPWAN standards, and things are getting worse rather than better and I think it is a terrible thing. We need a dominant short range and long range IoT standard. It is unfortunate that we haven’t got a dominant standard at the start that everyone can adopt, add their IPR and make money.’

Webb’s view is that while the proprietary standards may have first-mover advantage and are clearly fulfilling a need, they are unlikely to survive long term. ‘If you look at the history of wireless, the only successful technologies are open standard. There are no examples of proprietary ones being very successful, so I can’t see how the proprietary LPWAN vendors can last long term unless they morph into open standards.’

He points out that this is exactly what has happened with Weightless. UK firm NWave Technologies joined in October 2014 and Taiwanese firm M²Communication (M²COMM) in July 2015, with both deciding to make their IPR available to all by joining Weightless. ‘In effect they said: we won’t get this to work if we stay proprietary, so we’ll gift our technology to an open standard,’ says Webb.

M²COMM developed its proprietary Platanus protocol largely to target high densities of end devices (up to 10,000) in short-range environments, such as thousands of RFID tags in retail spaces.

Fabien Petitgrand, a member of the technical staff at M²COMM, explains: ‘We developed a pretty closed and proprietary application, but we wanted to go beyond that and move to longer-range outdoor applications for smart cities and so on. But we needed to add some kind of standard for interoperability, so we got in touch with Weightless SIG.

‘The benefit for us is that there are more parties involved, so you design a better standard and we can still be first to market with hardware and software, because we created it and so get a first-mover advantage.’

Petitgrand explains that the key consideration with LPWANs is trade-off between transmit power, range, the amount of data being sent and the interference noise levels in the particular spectrum bands being used.

‘For us it is really about capacity; the ability to handle the growing number of devices needing to communicate some data on the uplink. You need to get as many users as possible in one hour on the uplink. We are not seeing capacity issues in M2M yet as there is very little deployment, but we know from the cellular experience that this will become the main problem.’

He continues: ‘In unlicensed bands you also have to deal with other users you cannot control. Some will be narrow band and some wide band, but what it means is your noise level cannot be controlled as other users will be using the same spectrum. That means your range gets hit and therefore you cannot guarantee the range in those circumstances.

‘We are looking at narrow band, rather than ultra narrow band to try to guarantee better reliability. I don’t think the 20db increase in coverage aspiration is the point. It is not about achieving x km more. We are shooting for capacity and reliable communications, rather than trying to hit the longer range, so that is what Weightless-P is mostly about.’

Petitgrand says M²COMM looked at cellular options, but observes: ‘It is a longer path via 3GPP to reach licensed IoT LPWAN standards. Of course it will happen and we will play in that space for sure and we are designing Weightless-P in that spirit.’

Multiple options required?

Ovum’s Jamie Moss is less worried by the proliferation of LPWAN standards. ‘It is not a case of either having to use LTE or a narrow band option; it could be anything in between. What we want is to have everything we need to supply the market at large. Some standards may be too narrow band for some applications, but if we have a granular availability of categories at the right price point that will give the market a wide range of possible options.

‘That way we will have a full family of standards all moving towards 5G, which will hopefully have IoT considerations embedded in from the word go, so it can serve any particular use case from very high throughput with very low latency to very small throughputs where latency isn’t an issue.

‘It’s all about providing effectiveness and efficiency,’ argues Moss, ‘so if an enterprise wants to improve its processes to be more effective, more efficient and reduce its costs, it has a solution available. If the solution does not enable this to happen there is no business case.’

Olivier Beaujard, vice president market development at Sierra Wireless, sits somewhere in between. ‘There are multiple candidates for LPWAN, but we do not believe only one technology will dominate as there are so many different applications, so you’ll never get one that works well for everything.’

Sierra Wireless is not working with the proprietary independent providers, however. ‘We work with the different working groups in 3GPP and GSMA so after we can share the same alignment on standardisation. We do believe that IoT has to be standardised if we want a massive spread across different industries.

‘But we think there will be more than one winner. We are following all of the options in the standardisation process, but where we are pushing the most and guiding our road map is the LTE-MTC family. We are contributing to the others too – some just to see what is happening, as we may need to react to more than one family of products.’

He says that Sierra is more sceptical about investing in the narrow band options such as Sigfox. ‘We see them more as proprietary solutions and the disadvantage they have is that you have to deploy a completely new network, while the LTE-MTC family means you can reuse existing infrastructure. That said: we could have partners who might develop partnerships with them combining cellular and narrow band options.’

Beaujard adds that Sierra is not concerned that LTE-O won’t be ready until 2017 and LTE-M until 2018, as the company believes the key verticals it is targeting, such as the utilities market, are not ready either.

‘It is a long process for them, so we are not late for the type of market we are addressing. You can start with LTE Cat-1 and then move to Cat-O. We use the same form factors for our modules for 2G, 3G and 4G and we will do the same for whatever low-power solution emerges: that’s very important for migration,’ says Beaujard. ‘We need the MNO networks in place too. It is a roadmap alignment for them, so we need to make sure we are not too early or too late for their alignment.’

Ovum’s Jamie Moss points out that LTE was created for high bandwidth requirements, but argues that doesn’t mean it is all it is suited for. ‘If something does require a lower throughput rate you take existing technology, which will be around for 20 years plus, and adapt it; that’s an established pattern.’

It’s a pattern followed by the big M2M module providers Cinterion, Telit and Sierra Wireless, although Moss says they are likely to face increased competition from the likes of Huawei and ZTE, who are keen to break into this market.

‘They are going down the route of providing the cheapest possible module shipped in the greatest possible volumes to achieve economies of scale; they are talking about shipping in the millions and are earnestly pursuing this market. Huawei is developing its own modules and chipsets,’ points out Moss.

Who will build the networks?

The question of which LPWAN standard to use does rather come down to who is prepared to stump up cash to build IoT networks using the independent protocols available now. And then, what do the MNOs do? Wait for 3GPP to define LTE-M, so they can keep costs down by using their existing infrastructure and spectrum, or invest in UNB or NB alternatives in unlicensed spectrum now so as not to miss the boat?

Weightless SIG’s William Webb says: ‘Our initial assumption was that it would be the operators who would build IoT networks, as they already have the masts, management and back office systems. But having dealt with them for a long time I’m not sure about that any more. They are big businesses focused on delivering services to consumers, and moving from a business that makes billions from consumer services to IoT that might only make them millions may not be that attractive.

‘It might happen though,’ he concedes, ‘but my take is that we will see other companies deliver IoT networks first, and if they are successful and make traction, then the MNOs might buy them up. If you look now it is the likes of Arqiva, Sigfox and BT too, of course, that will probably lead the way.

‘We need something to drive it forward with some companies to take the early pain because they can see the benefit of doing that. It might be a commercial organisation, but I don’t quite see who yet. Or it might be some other entity such as a government wanting to drive smart cities – and that’s not a bad place to start,’ says Webb.

Ovum’s Moss thinks the MNOs may play a bigger part than some anticipate. ‘The thing is people think the MNOs are just one-trick ponies with cellular, but they are service providers. They want to supply everyone, so they must be as flexible as possible and offer everything that is necessary. If they appreciate there is a level of demand for something they don’t have, they might roll out completely different networks.’

In fact, in September T-Mobile announced it is going to roll out a Sigfox network across the whole of the Czech Republic, while Orange announced plans to implement a LoRaWAN across metropolitan France to complement its cellular M2M/IoT offering, so we may see the MNOs leading the LPWAN charge after all.

Webb says: ‘We are moving to a world where people are starting to make these kinds of investments, but it is a hard one to call as to who and what will emerge on the back of what applications. It is quite a risk.’

Image Credit: Melpomene / Shutterstock

• What is IoT 2.0? Plus why you should care