There are still many challenges to overcome before the Internet of Things (IoT) becomes a widespread reality. Not the least of which is trying to establish a general agreement as to what it is and how it differs from machine-to-machine (M2M) applications.
Robin Duke-Woolley, CEO of Beecham Research, offers a definition. ‘The general difference between the two is that M2M tends to be vertically aligned within one particular industry or business process, while IoT is horizontally aligned across a number of sectors with different values being given to the data depending on the applications it is used for.
‘The purpose of a smart city should be better living for citizens. That is a distinctly different outcome to a typical M2M application where you are looking for improved business processes,’ he says.
‘A lot of people talk of M2M as a subset of IoT. You can do that, but you need to separate out vertical and horizontal applications, because they have different technologies and service issues.’
Duke-Woolley points out that for IoT applications to really take off they have to be assessed from both a technology and a business model perspective. Is it technologically possible and if so what is the best solution? And does it make sense as a business proposition?
‘A typical M2M example is fleet management: where are my trucks? The data is used for a specific purpose and it is cost justified for that application,’ he says. ‘With IoT, if you just collect a piece of data from a soap dispenser or a toaster, it probably won’t be a cost effective exercise in itself. But if that data could be useful elsewhere, it might make business sense to collect it.’
For example, if a connected toaster is used in the morning by an elderly or infirm person living alone, that information can be used to alert care services or family members. An energy company might use the information for predictive energy usage or the building security company to confirm someone is occupying the apartment. This multiplicity of users will help cost justify the application.
IoT is likely to develop in waves, according to Duke-Woolley, led by the most obviously cost justified applications.
But we need to get better at mixing and matching data and enabling the machines to find and recognise each other. We also need to address how to extract meaningful value from the vastly increased amount of data being collected.
‘We need to think about how the infrastructure for IoT is designed and used – and there is not a simple answer. A lot of work needs to be done on databases to enable that,’ says Duke-Woolley. ‘And how cost effective is the connectivity at the moment? I think it will be a long time before we get to the connected soap dispenser.’
To ensure IoT becomes a reality, work still has to be done in all parts of the IoT stack starting with the edge technology layer (IoT modules/devices) up through the access gateway layer, internet layer, middleware layer and finally to the applications layer.
IP to the edge
Starting with the edge layer, Pilgrim Beart, CEO of UK start-up 1248 Ltd, which has products in all parts of the IoT stack, says: ‘IP to the edge has not really been done yet and that is what has locked us into the M2M world.
‘This is because the gateways connecting the edge to the internet are proprietary, so there is no standard way to translate the edge to the internet.’
Beart says IoT gateways need to be like Wi-Fi access points where all the manufacturers’ products are designed to the same standard. ‘You need openness and interchangeable layers of IP across the IoT stack to achieve that horizontal vision.’
Looking at the IoT challenges that have been nearly overcome now, Beart says IP to the edge, bandwidth issues, low power consumption, long battery life, more powerful processors, miniaturisation of devices and sensors have mostly been solved, all of which is helping to drive down costs to the point where IoT becomes affordable.
Philippe Guillemette, chief technology officer at Sierra Wireless, one of the largest IoT/M2M module manufacturers, agrees but says the industry needs to make it easier for application developers to design products.
In response to this need, Sierra Wireless unveiled its Legato app development kit at Mobile World Congress this year. It is a Linux based, open source kit designed to allow customers to create their own IoT applications from an easy to use template.
Guillemette notes that some IoT applications will require two-way communications and will often need to be capable of some processing at the edge rather than sending everything back to the centre.
He says: ‘You can design a device that can do edge processing or is capable of gathering data to do some analysis and then send it to the cloud. But you need a strong CPU to do edge processing, so what we want to do is build all the horizontal pieces necessary to make it easy to develop IoT applications.’
Device sleep mode
Another challenge that has had to be sorted is the fact that edge devices are not only low power, but are asleep most of the time to save energy. Beart explains that if you need to communicate with a device that is asleep, the edge gateway which is ‘on’ all the time, can act as a proxy for the device. If the device needs to switch on a sprinkler or change a parameter, but is in sleep mode when the instruction is sent, the gateway stores the instruction and communicates it when the device is ready.
‘Gateways have always done this,’ acknowledges Beart, ‘but they were all propriety before, so they did it in different ways. To get round this a protocol called CoAP (constrained application protocol) was designed, largely by the IETF, to solve this in a standard way.
‘For example, if you have an app that wants to know if the temperature changes, you don’t want it polling all the time. CoAP registers a call back instruction at the gateway; it allows web clients to observe or enables them to subscribe to changes to the device. It is a very efficient way to hook web apps to edge devices.’
However, once the data has been sent to the server the problem becomes how to discover it. Applications are written using a specific API to retrieve data and put data on them, but to have a human trying to write code to make every IoT application work with an IoT service it’s never seen before would be an impossible task.
This is where solutions like the recent UK innovation called HyperCat come in. HyperCat was developed by a consortium of 40 UK-based technology companies (including 1248) funded by the UK’s innovation agency, the Technology Strategy Board.
The aim of HyperCat is to provide an easy way for similar data sets to discover each other on the internet without humans having to write programmes. This will help to break down vertical data silos and provide a foundation for connected products and applications to interoperate.
Beart explains: ‘HyperCat asks what is in the data service and allows you to go and look for other data it understands on the internet, such as applications measured in degrees centigrade.’
HyperCat gets round the vexed question of ontologies or taxonomies on the internet by discovering data and tagging it, but what tag you use is up to you. In this way, you can create your own ontology or just adopt someone else’s.
Up to now wide area IoT applications have largely been forced to use cellular networks to transmit data from the edge to servers simply because there has been no other option available.
There is no doubt that 2G, 3G and 4G networks suit some applications, particularly for M2M solutions demanding high bandwidth such as video. But many argue cellular technology, as it currently exists, is unsuitable for many potential IoT applications – and not purely on cost grounds. New standards such as Weightless may prove a better option. So, how far are we from having wide area connectivity suitable for IoT?
‘We are not there yet and we are still a fair way off sadly,’ answers Professor William Webb, CEO of the Weightless Special Interest Group (SIG). ‘I like to apply the Chinese washing machine test. If you are a Chinese washing machine manufacturer and you want to provide a connected washing machine to add new features for your end users, what connectivity solution do you choose?
‘You could opt for 2G, 3G, 4G, Bluetooth, ZigBee, Wi-Fi or Weightless, but which should you choose? You conclude that you don’t know and so you can’t do anything yet. We’ll know when IoT is properly established when the Chinese washing machine manufacturer can say: of course, my connectivity solution has to be this.’
Webb says that if all you want is a local area or short range communication solution then options like Bluetooth or Wi-Fi, as used by the Nest thermostat, for home technology are fine. But if you want wide area IoT communications for Smart Cities, the only two real options at the moment are Weightless and LTE-M (LTE Machine to Machine).
The Weightless standard was designed specifically as a new wide-area wireless networking technology for IoT, aimed at going beyond the reach of today’s cellular solutions in terms of coverage, battery life, module cost and efficiency goals.
Weightless still needs a semi-conductor manufacturer to develop the chip sets, but Webb hopes they will appear in the next 6-12 months. The standard is a cellular style network with a range of up to 5km. The idea is that an operator (not necessarily an MNO) would put in a network of Weightless base stations stretching across the UK.
‘The devices communicate direct to the Weightless base station, so the key part is getting someone to build a network,’ says Webb. ‘BT is building a small test network in Milton Keynes, which will allow people to build solutions.’
The SIG calculates that a Weightless network would be one third the size of the typical UK MNO network (c.15,000 base stations) so it only needs about 5,000 base stations - not a big deal compared with many other installations. End users would buy a Weightless device, sign up with the operator and be off and running.
The other potential wide area connectivity option is LTE-M which is being worked on by the 3GPP standards body. Currently, LTE is far too expensive for most IoT applications. The aim behind LTE-M is to find a cost effective way of using 4G spectrum to make it suitable for low power IoT applications.
The first major development, due out this year, will enable many more low power IoT devices to use the same LTE spectrum. ‘These are called category zero LTE devices,’ says Sierra Wireless’ Guillemette. ‘They don’t care about providing lots of Megabytes per second, they are designed to do the job efficiently and cheaply enough to be attractive to the market.’
The next step, due out a year from now in LTE Release 13, has two ‘very important elements’. The first relates to power consumption and is about changes in the ways IoT devices sleep. Here, the objective is to reach the energy budget of AA batteries, which should multiply by 100 times the lifetime of cellular IoT devices to provide years of operation on very simple batteries. The other element coming down the LTE-M pipeline is a 15-20dB improvement in coverage.
‘If you put these three aspects together for LTE-M you get improved cost, coverage and battery lifetime. But it is also LTE based and that leverages a gigantic infrastructure. With these changes the power, connectivity link and cost become compatible with deploying a much larger number of devices. It takes time to achieve standardisation, however, but once done it will have massive benefits,’ believes Guillemette. However, others advocate that 5G might be a better standard for IoT and 3GPP should concentrate on developing other aspects of 4G for now.
Whatever the outcome of that argument, Webb observes: ‘The LTE-M group still has to write the standard, evolve the chips sets and get them out there into the devices, so it is probably five years away still. The problem is we need something now, so that suggests other standards will be used at least between now and a cellular solution appearing. That leaves Weightless as the one wide area standard ready now.’
Naming and identification
Another major IoT challenge is that each IoT device will need to have a unique address otherwise you can’t know whether you have contacted the right machine. So, you either need a general address system that everyone can use like IPv6, or you opt for a more closed address system – it will depend on the application.
‘Take smart meters,’ says Webb. ‘npower gives a name to each smart meter in any way they like, which integrates with their database from Oracle or SAP. The fact that that database is not accessible by a power company in South Korea is not a bad thing – so that naming approach for IoT devices may be fine for many applications. But if you want to make available a data resource that anyone can access, like a street temperature gauge, you might need to adopt a more general numbering scheme.’
IoT security continues to raise major concerns, especially as many solutions only address part of the stack. ‘Security is the big white elephant in the room,’ agrees Duke-Woolley, who adds that no one provides an end-to-end security solution yet.
‘At Beecham Research we are looking at an opportunity to create road maps for IoT security that go across different parts of the value chain. Each stage adds a piece, as we need to have security solutions that snap together at different points. But we have to start with the semi-conductor designers. You cannot just resort to software for security; it has to be there in the silicon right from the start,’ insists Duke-Woolley.
Privacy and trust issues are also becoming more of a challenge. Who has access to our data and who controls it? Beart says: ‘Privacy is a much, much harder problem to solve as you are dealing with peoples’ rights and expectations. Usually, it is a trade off between usability and getting value out of something and peoples’ privacy. The key principle here is openness and transparency; people should know what is going on.’
These social, legal and cultural issues will all need addressing as IoT becomes more prevalent. Duke-Woolley points out: ‘We may solve the technical side of IoT, but there are a lot of individual preferences involved and we cannot assume there will only be one solution. With some IoT applications we can say: this is what we could do technically, but we must also ask, do we really want to do it – do we in fact need to do it?
‘IoT ideas are put forward primarily by technologists,’ he continues, ‘but do people actually want them? There needs to be a conversation between suppliers and users as to what is useful and what they are comfortable with.’
In Beart’s view the main qualitative challenge is the sheer diffuseness of IoT. ‘There are lots of people working and waiting for norms to emerge with IoT, but there is a lot of work to be done yet to identify what those norms are or will be.’
Webb believes IoT could change the world by making things more efficient in a way the next generation of cellular won’t. ‘But because it is something that touches all the elements of our lives it is difficult to get all the necessary elements corralled and correctly aligned to enable it to take off,’ he says.
‘It is complicated, but history shows it does happen, it just takes a while to get there. We still need that Apple app store/iPhone moment for IoT and then we need a solid, ubiquitous, secure network that passes the Chinese washing machine IoT connectivity test,’ says Webb.