Battle lines are being drawn between the standards that want to bring sensors out of the field onto the Internet of Things.
For all the talk of the Internet of Things (IoT), most things will never really connect to the worldwide network. They will sit on an intranet and talk to the cloud through a gateway that sits mere metres away. But there will still be plenty that need their own direct connection because they will be too far away to make installing a gateway viable. Whether sitting on street lights or inside the level meter for a reservoir, they need access to a low-power wide-area (LPWA) wireless network with a range measured in kilometres.
Until recently, the main option for organisations was 2G cellular. Vending machines and other units that are often left out in the open with no realistic access to an Internet gateway have used the GSM Packet Radio Service (GPRS) for years. But GPRS does not look to be a good long-term candidate for use on the IoT. Operators are now looking to reuse GSM’s spectrum to provide more capacity for their 4G networks.
Wiren Perera, senior director of IoT strategy at On Semiconductor, says: “What we see emerging as being different are very low-power and fairly simple cost-effective sensor networks. Cost-effective has two sides to it: one is the cost of the bill of materials for each device. There is also the operational side of it.”
One problem that the existing GSM network has is that, originally designed to handle voice traffic, it was never intended to cope with huge numbers of devices that expect to to be attached to a base station for long periods but send and receive very little data from day to day. The new crop of networks popping up to supplant it were and they include stripped-down versions of 4G cellular that should cope better with a growing population of smart street lights, utility meters and buses.
Some meters present a problem for wireless networks. Sensors that detect parked cars usually need to be buried. Water meters go even further underground, although they might not be entirely surrounded by concrete and asphalt.
Emmanuel Gresset, business development director at digital signal processor (DSP) specialist Ceva, says: “The deeply buried water meter application is probably one of the most stringent from a coverage standpoint.”
Stefanie Giesing, head of business operations for IoT wireless operator Digimondo, says: “The company was started by E.on employees last summer. We were searching for solutions to connect smart meters that are often located in cellars. For us it was really important to find a technology that could reach nearly every meter in every location.
“In one location near Hamburg there is a swimming bath with 40cm-thick walls and no windows. It was horrible for mobile connections.”
The team found the LoRa network protocol developed and licensed by chipmaker Semtech best suited their needs. “But we had the problem in Germany that no-one provides a LoRa IoT network. Telecom operators are a bit quiet when talking about IoT networks. At the moment it’s more theory than practice. So we decided to do it ourselves,” Giesing explains.
Digimondo started rolling out networks in Hamburg and Berlin, adding Nuremberg over the winter to support a demonstration of IoT technologies that coincided with the Embedded World show there in February. The plan is to stick to cities in the short term. “But when we think about the metering case we will also go into more rural areas,” Giesing adds.
For rural coverage, cellular still has an advantage in terms of the base stations that are now in place but it has taken time for the industry to put together an attractive offering for IoT-focused networks. The first step is Cat-M, a variant of the protocol used in LTE handsets. Although it strips out some of the complex hardware that a smartphone handset needs, Cat-M is still hungrier for digital signal processor (DSP) performance than a protocol like LoRa.
Gresset says: “To be power-efficient you will need hardware accelerators alongside the DSP in a Cat-M design.”
But the cellular industry is not pinning its hopes on Cat-M. Although it has taken time to get off the ground, the 3GPP group that defines cellular standards such as LTE and the upcoming 5G portfolio is working on a simpler form of 4G wireless, Narrowband-?IoT, which has more in common with a simpler protocol like LoRa.
“We have been seeing a very big push on Narrowband-IoT lately. It seems to be speeding up,” Gresset says. “It started in December with some of the first meetings by 3GPP. There is a convergence on Narrowband-?IoT that will replace GSM from a cost standpoint and probably even get to a lower price point.”
Gresset says a Narrowband-IoT interface is likely to need half the silicon area of one aimed at Cat-M because of its lower complexity. The lower complexity will help the kind of silicon integration that can bring costs down further for companies that make chips for meters and IoT nodes. “There is the possibility to have the RF integrated and potentially even the power amplifier and flash memory.”
Although there is nothing preventing similar integration for LoRa, Semtech’s role as a silicon vendor in control of the core technology means there is less opportunity at the moment for other companies to do the same kind of integration that will be possible for Narrowband-IoT or Cat-M. Semtech says it is looking at other licence deals in addition to those it has signed with Microchip Technology and STMicroelectronics but that they are assessed case by case.
Perera argues that the Sigfox protocol can achieve lower cost both in terms of the silicon because of its relative simplicity and operational cost. “The modules are very cost-?effective. If you contrast with narrowband IoT, Sigfox is likely to be two to three times lower on the module cost.”
In contrast to LoRa, Sigfox has a single operator which has embarked on a rollout programme in various parts of the world. To improve adoption, Sigfox is much happier to license the transceiver design so that controllers can be embedded in custom silicon and cut implementation costs. One big difference from the other protocols is that Sigfox only transmits in one direction and at very low speeds, potentially over distances of up to 15km.
One-way communications is an approach that EnOcean adopted for shorter-range communications in devices such as wireless light switches. In principle, it can reduce overall energy consumption because the IoT node does not have to listen for acknowledgements. But there is no way for the sender to know that its messages are being received. And, unless there is a reverse channel, no way to deliver updates that may be needed to maintain security. Sigfox might be used in concert with a protocol like Narrowband-IoT, with the cellular channel being used for software updates and device management, which can take place much less frequently and so reduce stress on the battery.
“The question there has to be: do you need the richer traffic and then do you want to pay for it?” Perera asks. “We see Narrowband-IoT almost as being complementary and different and still probably two to three years out.
“Another one that is coming up is 802.11ah, a form of Wi-Fi. It’s different to Wi-Fi in that it’s in the sub-gigahertz range,” Perera says, contrasting it to conventional Wi-Fi that uses either the 2.4GHz or 5GHz bands. “To some extent, it addresses the range issue of Wi-Fi. It is probably not as low-power as the other technologies. You can get to hundreds of milliamps so it is significantly lower than traditional Wi-Fi but not in the same range as the products we offer today. But it is a higher bit rate that’s similar to Narrowband-IoT and so could bring a value-add.”
The experience with Wi-Fi for conventional computer networks shows that a consideration for users, alongside silicon and operational costs, is quality of service. The non-3GPP products operate in unlicensed spectrum, which means they will compete with other types of radio user. LoRa attempts to deal with interference using a form of encoding that is designed to spread the RF energy across a large slice of spectrum and so avoid being jammed by other narrowband users, which could include Sigfox.
White-space radio promoted by start-up Neul expanded on the idea of interference avoidance used by Bluetooth. Each RF device sniffs the airwaves to find unused white space in the spectrum that it can use to communicate. UK regulator Ofcom was relatively quick to explore the possibility, allocating spectrum provisionally to white-space radio in the old analogue TV bands. Other regulators have not followed suit, making it difficult for the technology to gain traction.
Perera says: “The challenge that we see with white space is that the technology is OK in the band it operates at and with the range you get. The challenge is that it’s in the TV space so it’s very stringent from a technical standpoint and it also needs regulatory change.”
Tony Milbourn, vice president of strategy at module maker U-blox, says: “The hazard for the customer arises if something affects the link performance, such as wireless microphones that are being used for concert nearby. If they scramble the channel maybe you can’t read the level of the reservoir. In licensed spectrum that will not happen,”
Because it is likely to take several years for Narrowband-IoT to be implemented in both devices and networks, a lot of uncertainty surrounds the LPWA market. The wait for a cellular standard purpose-built for the IoT provides Sigfox and LoRa with an opportunity to grab market share before the major operators move in. Cities and infrastructure companies can build their own LoRa networks, which will increase competition and put downward pressure on the cost of accessing the Internet.
The current situation means device makers are faced with a difficult choice as to which to adopt. They may simply decide to go with several in the hope that, eventually, they can settle on one. But in the short term that will increase device and operational costs - and potentially slow down the creation of the long-range IoT.