Nordic Semiconductor is playing a key role in the future of electronics by providing ultra-low power (ULP) wireless chips that can run for long periods of time from power sources as small as watch batteries. But that’s just the beginning.
We spoke with Thomas Embla Bonnerud, Director of Product Management at Nordic, to discuss the company’s approach to the Internet of Things. In a free-ranging conversation, he covered everything from Bluetooth Smart to wearables to phones as routers and whether or not routers have to be smart. He also gave some great examples of industrial uses for the Internet of Things. Suffice it to say, your refrigerator and coffee machine may never break down again.
Please share with us a little bit of the Nordic story and how the company decided to aggressively go after the Internet of Things?
One of the things that we at Nordic understood after being one of the early vendors of Bluetooth Smart was that a lot of people initially thought that Bluetooth Smart is like – Bluetooth – you know, it’s cell phone accessories. It’s like, now you can do keyboards, now you can do proximity tags and you can do heart rate belts – you can look at your heart rate on your phone.
One of the things we realized is that yeah, that happens, but the real driver we saw was that the phone was typically used as a gateway. So, for example, if you look at wearables today, what’s really going on is that Bluetooth Smart is using the phone as a router to connect your cloud service. The real value of these products that we were in were actually a pairing between a thing – in this case, the wearable – and then the cloud, which is the cloud service. Try to imagine Fitbit without Fitbit.com, the application, the stuff that is happening in the cloud.
And think about the same thing with Garmin, right? So we sort of realized very early that what’s actually happening with Bluetooth Smart is that it is connecting my personal things. Things that I wear or that I have, but it’s near my cell phone, to the cloud, and then the phone basically acts as a router and it is the app that you install on the phone that enables that device – that app will essentially bridge from Bluetooth to IP and up to the cloud. So we looked at what we had on the market and said, “Okay, this is a nice solution. It’s a very, very nice solution for my things – my personal thing. My watch or my wearable and whatever personal devices that I carry around. It’s a nice solution to be able to connect to the internet while on my phone.”
But then after looking at this, we started to look at the bigger opportunity, which we think is connecting the things around us. Not only my things, like the wearable, but what about connecting a board game to the Internet, for example. Or my light bulb to the Internet, or my coffee machine to the Internet. And when you think about those things, they are different because they are typically not personal things. Multiple people share access to these things. And one can also say that in many cases, you may want them to be connected 24/7 independently of whether someone is around with their smartphone or not.
So if you think about smart home, connected industry, retail buildings, and maybe machine to machine, you start to thinking, “Okay, maybe there’s an opportunity for Bluetooth there.” It’s low power. It’s much more mature. It’s an open standard. There is a wide variety of chipsets available, and it’s a fast-evolving technology, adding new features along the way. So why can’t we take these applications?
And one of the things you sort of realize looking at these things is that the phone as a router simply doesn’t cut it, right, because you can’t keep a phone around and install in an app. That’s where, really, the whole thing started and we saw the need for a different solution and that solution was, for us, pretty obvious to be IPv6 over Bluetooth Smart and that was aligned with some of the stuff that was happening in the Bluetooth SIG, but it was also aligned with similar industry initiatives that were already moving at this point in time.
You have 6LoWPAN on 15.4 I think even at that stage Thread [the technology used by Nest and other smart home devices] started to appear and Thread is essentially IP over 15.4. There are always people that argue that Internet Protocols are so heavy. If you want to do low power, if you want to do low cost, there is no way to rev IP. We thought differently, if you look at the capabilities of the chips we had back then – especially if you look at the new chips we have – running IP on these chips is not a problem at all.
“This idea that you are going to have an Internet of Things that is architected and runs in a different way than the core Internet, we believe, is a very stupid idea.”
This whole idea is that in order to connect things it has to be so simple. And our approach was simply that we take the IP all the way to Bluetooth mode, and each Bluetooth mode has an IPv6 address. This aligns Bluetooth with Wi-Fi, Ethernet, and other technologies that are being used across the Internet today. It also means that instead of reinventing security protocols or transport protocols, you leverage things that already exist. We definitely believe that once you want to take the step from my things, for the phone and standard Bluetooth, to leverage Bluetooth in a bigger scale, in static and distributed networks, IP is the way to go. We think it’s the future-oriented solution.
We have a couple of key points, why we think this is the case and the first one is pretty blunt, and it basically goes like this: You want to align with the Internet today. This idea that you are going to have an Internet of Things that is architected and runs in a different way than the core Internet, we believe, is a very stupid idea.
There is one proven, massive network of devices today and it’s all based on IP. And it applies to the Internet principle where you have smart end points – the device itself and the servers – and the network is dumb. So, you need smart things and smart services. If you look at the Internet today, you don’t – in the massive scale of Internet – you don’t have these smart gateways that do a lot of translation and so on. Everything is essentially dumb, and then all the magic happens over IP. The Internet and that architecture is proven – it’s proven to be scalable, it’s predictable and it’s reliable. Once you want to plug Bluetooth into the Internet, you don’t want to reinvent the wheel, you only use proven Internet and technology.
Number two, related to that is what we call, enabling dumb edge routers. If you run IP on the Bluetooth node and you don’t need a router that has to do any sort of prep related to translation or have special security protocols adopted, you essentially discover the Bluetooth node, and set up a connection – a data link – the way you do Wi-Fi, the way you do Ethernet. And then the edge router is not smart. It’s simply running IP addressing. And forwarding packages one way or the other. And again, we think that these dumb edge routers are key to scalability and key to interoperability. Otherwise, you have to match the edge routers with the things, or even worse, with the cloud services. And that’s going to happen.
We’re going to see gateways. We’re going to see gateways tied to things. We’re going to see gateways tied to cloud services. But we think, ultimately, if it’s going to be by the billions. It can’t be that way. It needs to be a separate play on the things, a separate play on the edge routers, and a separate play on the cloud services. They need to work together the same way that you can go into a store and buy a Wi-Fi router and there is no special application you have to install on the Wi-Fi router to be able to talk to your PC or your phone.
The interesting thing is that on the router side of things are pretty far along. If you look at OpenWRT and Bluez they support Bluetooth 6LoWPAN over Bluetooth. So we’ve proven that with some simple hacks to existing code that is running on available Wi-Fi routers today, plug in a USB router into Bluetooth Smart, and you have an IPv6 over Bluetooth Smart router.
So, our existing conventional home routers can be upgraded?
Yeah, of course, the router needs to have a radio, and some routers have that today. If you look at the newly announced Google router, it has support for Bluetooth Smart. I don’t know exactly how Google is going to play it, but if you look at some popular operating systems that are typically used for routers – one of them is OpenWrt. So, a lot of commercial routers are from that operating system, that operating system already has support for IPv6 over Bluetooth Smart. So it’s just a matter of the chipset that’s in there to have Bluetooth Smart.
We think the barrier is pretty low because a lot of the existing routers today – if you look at those Wi-Fi chips – they are typically combined Bluetooth and Wi-Fi chips. The sort of threshold on the router to enable Bluetooth Smart is relatively low and as long as it’s IP it doesn’t have to be smart, or have intelligence. On the router side, it’ll handle IP packages for Bluetooth, the same way that it’ll handle cycle packages for Ethernet or Wi-Fi.
What do you think the timeline looks like for us to see this actually happen?
To see routers with support for Bluetooth Smart, or IPv6 over Bluetooth Smart? I think that we are going to see the first solutions this year.
What I should be clear on is that even though you have a specified way to run IP on Bluetooth and you can match that on the router side — there are still things to do on the standardization side regarding commissioning and joining that works and so on, which is still happening. So I think that it will still take some time. I believe a lot of things will be happening this year on aligning these technologies and making sure that the out of the box experience is good. I think that’s something that will come along.
Regarding the new Google router, OnHub, do you think that helps? Does that move things along and did you guys have any technology contributions to that?
Well, the only thing I know about that one is that what Google says right now is that 15.4 and Bluetooth are not enabled yet. So they say it’s 15.4 and Bluetooth requisite. Probably has the radios and the antennas. But I would suspect that Google is working on the software supporting the router. And I don’t want to speculate exactly how they’re going to enable it, or whether it’s going to be IP. But I think that if you look at Google’s contributions on short-range wireless, they have been very IP focused. I hope and my best guess would be that it is, but I don’t have any information.
“We’re… getting sensor data to the cloud to figure out what is the state of this thing, when is it going to break, when do we need to service it at the right point, and can we then order the parts online and have everything ready to go?”
You’re kind of alluding to other things. Could you tell us a little bit about mesh networks and Bluetooth? We had also heard that the SIG chairman says mesh networking is coming to Bluetooth. Can you give me a layman’s version of how that works and why that matters to the IoT, or does it?
Right so, it’s important to keep a distinction on IP and mesh networks because essentially the mesh networks are the data link layer or the flexibility of the data link layer. And a mesh network could carry IP packets or it could carry non-IP packets. So they’re sort of two different but definitely complementary technologies. We believe that mesh is an important technology in the context of Internet of things because it could enable you to have a larger number of small and battery-powered devices that use each other to expand the range.
For example, in a hotel you have connected door locks, right? And on the whole floor, instead of having one base station that has ranged to reach every door lock in that floor, they could be connected together in a mesh. And they could use that to hop between door locks to get to the base station and then ultimately to the power source. So, I think it’s a key technology for those larger installations and especially installations like door locks, lightings or thermostats where you have relatively close by nodes and there is not an extremely lot of activity so that you can afford the latency to make the jump between them.
The alternative to a mesh network is a star topology where every node needs to be within range of the gateway or the router, which means that either you have to have higher density routers, which are more expensive devices, or every one of these nodes and the routers need to have a longer link projection, which typically has a price tag in terms of silicon and a price tag in terms of power absorption.
Now, that said, I think that mesh is not something that is necessary because if you think about it, all of the large-scale deployed network today — well, 99% of them — are star networks, cellular is a star network, WiFi is star network. So, either way, I think that a lot of things will sort of start with star topologies and as mesh and Bluetooth evolve that will sort of compliment and push to the limits of what you can do.
Beacons and security. My understanding is that the Apple iBeacons can be vulnerable in that their IDs can be duplicated. So hackers can potentially create “false” iBeacons. Whereas my understanding is that with the Gimbal Beacons, you cannot impersonate a Gimbal Beacon like you can an Apple iBeacon.
Yeah. Definitely we see people pushing the envelope on beacons. And one of the things they do, and they can do with our technology, is that they connect the beacon to the cloud. So instead of sort of like a dumb beacon that just sends a unique ID, and it sends that all the time, and all I have to do is to go and listen and I can recreate it – it can actually do rolling IDs. So it can sort of intelligently use different IDs at different points in time. Then when your phone comes around it gets this ID, sends it to the cloud, and it gets meaning out of that ID. And the cloud and the beacon are synchronized so the cloud would know that if this beacon sends this ID at this particular point in time, it means this. So that you don’t have this fixed relationship between a beacon sending a fixed address and that matches some sort of cloud response.
So yeah, definitely. IP could play a part in this. One can imagine beacons that are networked that have IP addresses that can be then safely reconfigured from a cloud service using SMTP or a web protocol to update the beacon address. And then at the same time they actually beacon, they send beacons.
We’ve touched on wearables and the consumer side, but are you doing much on the industrial side or rather the Industrial Internet of Things (IIoT)?
Yeah, definitely. We are. With this solution and other solutions, we are seeing a pull from the industrial side. Part of the objective behind the idea of IPv6 over Bluetooth is actually on the industrial side because we see that a lot of industrial systems are already based on IP and they have local servers running web applications and they want to expand these networks, for example, with Bluetooth. And running IP over Bluetooth enables them to create what we call heterogeneous networks, because if you have a router that supports Bluetooth, after that router, the actual intelligence, the server, has no idea or doesn’t have to know that this device is Bluetooth, or whether it’s WiFi, or whether it’s internet.
It allows them to keep the upper layers of their infrastructure. They keep the same services they built, and they plug these things in as standard IP devices. Especially in the industrial space, it is very important for them to line up layers and most of them have opted, and they’ve opted on IP for the upper layers.
Have you identified any hot-spots in the Industrial IoT? What do you see on your side? Is it sensors? Is it, maybe factoring? Is it logistics? What are the most interesting applications that you’ve seen?
The most interesting application we see at the moment or, let’s say use case, that we see at the moment, is predictive maintenance. You have a device, or a machine, or an installation, where you spread out or you have sensors, that essentially do sensor readings. They either aggregate locally or they send the data via an edge router to the cloud. In the cloud, you have machine learning to figure out: Is this thing going to break soon and potentially what is going to break?
So let’s use a stupid example. Let’s say my refrigerator – just to give you an example not related to a customer. You could probably put a simple temperature sensor and a vibration sensor in my freezer. So there’s a freezer downstairs in my house. And based on statistics, if you look at how the temperature varies and the vibration of the freezer, over time you can collect data so that you know that this freezer is going to break in three days and it’s probably this and that part that’s going to break.
That’s sort of the most interesting thing, and predictive maintenance is nice because there’s just massive potential to save money. I’m not sure you’ve seen the London tube thing. The London transport authorities have actually installed a lot of sensors in the London tube, to do exactly what I’m describing.
I’ve seen conversation around it changing business models – with product companies becoming service companies. As in your example, you have an elevator company where — now that they can do predictive maintenance — their business model changes in that they are now a service company or that they can preemptively sell repair parts or services. Do you see it that way?
Yeah, definitely. It allows companies to sell what were traditional products more as a service. It allows, as you say, an elevator company to not sell elevators with service contracts anymore. They could sell elevators as a service. By leveraging sensors and Internet technology, machine learning in the cloud and so on, [companies] can innovate and disrupt and do it significantly better than others. And the idea of this is that if you do this the right way, the cost of owning an elevator goes down.
“Imagine roads as a service…. You lay down the road, you sensor the state of the road, you fix it proactively, and you get back the roads. And those who pay for roads get more for their money.”
Imagine roads as a service. You know, you don’t pay one company to lay down the roads that has absolutely no interest in the quality. You lay down the road, you sensor the state of the road, you fix it proactively, and you get back the roads. And those who pay for roads get more for their money. And I think you will see this all the way down to commercial-grade appliances.
Think about the coffee machine. The frustration in our office when the coffee machine on the sixth floor breaks down. You have these people who want to get their morning coffee. Our company would be willing to pay or prefer a vendor that offers a coffee machine as a service. Where they say, “Okay, we are going to make sure that every morning when people go in, this coffee machine works. No fails, no calling, nothing. If something is about to break, we will call you, we will fix it and it will work the next morning.” This will happen, no question about it.
So, we are already seeing new business models emerging and you think things are advancing quickly?
Predicted maintenance as a technology and using multiple, different connectivity technology and sensor technology is now. It’s happening now. You already have big players doing it. So, it starts off with big, expensive stuff and then it’s going to trickle its way down. It’s hard for me to guess exactly when the coffee machine in our office is going to be cloud connected. But hey, it could happen very fast, because all of the technology elements are there. The short-range wireless, the sensor technology, the cloud platforms from Microsoft, Google, and others. It’s all there. It’s a matter of companies innovating and taking advantage of this technology to build it.
The Internet of Things series is brought to you by Treeline Interactive and is authored by Tim Homuth and Joe Austin. Our objective is to provide you with information about the rapidly evolving field of the Internet of Things by bringing you interviews and insight from industry leaders. If you have any questions, insight, or suggestions for articles please email us at IoT@treelineinteractive.com.