ARM and the Open Internet of Things | Internet of Things Forum | ARM | IoT

Simon Ford, mBed at ARM

ARM’s strategy for an Open IoT system and the terrifyingly long time-scales involved in building an industry.

Slides, Video & Transcript Below

Slides from Simon Ford’s talk at IoT Forum here

 

The BLN is now Business of Software

Business of Software runs conferences for people that build great software businesses and products. To access talks online, hear about new events, contact speakers and stay in touch with great ideas, share your email address with us.


We'll NEVER sell your email. Unsubscribe anytime.

Simon Ford, ARM:

OK, I’m here from ARM and I’m going to talk to you a little bit about some of the inevitability of internet of things and then perhaps a little bit about what we’re doing about that. What I decided to start with was I looked up on this website what I was going to be talking about and it told me this. That’s what this talk’s going to be about, but also someone had chosen a nice little biopic of me, which I didn’t recognize initially. It’s not one I had used. It got me thinking, this is actually a picture taken in 2008, so I haven’t aged too badly. I don’t know where this was found, but I guess someone googled me or something.

But this was me at a BBC hack. A 24 hour hack event called MASH. We were building hardware. This was an event where people were coming together to hack around with BBC APIs. They had just released a set of APIs for their platform. I think Yahoo had done the same.

The reason it was called Mashed was because it was about mashups. This was a new thing, so the new sexy thing you can mash up. People were going to expose their web APIs and you were going to mash these things together. Boardrooms were starting to wonder if this was something interesting.

We turned up with some hardware. So this is something that we were building that was, what’s on here, there’s like a GPS modem, screen, 3G modem, so on and so forth, gyros, an RFI motor. We built this thing over 24 hours. Just out of interest, what it did was you scanned a parcel. This was the idea. It was called Packet Network. You scanned the parcel and then you clicked on Google maps where you wanted it to go.

bitibitbitboopboopbipit

Now this thing sort of just went bitibitbitboopboopbipit coz it knew where it was, it knew where it wanted to go coz is talked on the Internet to find out where this packed was. It was kind of the idea that you picked up a parcel, you walk along, and when you are not going that way anymore you put it down. Then someone comes along and picks it up that way.

Anyway, bit of a stupid idea, but the we were doing hardware there, which was a bit of a novelty. And for me, was also quite insightful, because everyone was working very fast. We got stuck a few times and people around us helped us out like how to get onto the Vodafone network through some weird configuration or something.

It felt like we were cheating a bit. I was learning Google map APIs. They were only two years old at that time. This was a novelty. It felt like, OK I can get that to work. I can get this thing clicking and it worked fine.

We were doing hardware, so we were building these tools. We were trying experiment with what rapid prototypes I think for electronics would look like. We had these tools and that allowed us to get this hardware kind of working in really not very long.

We got this process working but if you’d asked about when I’d thought about what we’d actually built, I could kind of see how the web stuff could get to a production kind of thing in not that long. Yet, I felt like I was cheating on the and even the hardware. Yeah, get this on to a PC or something. I could see in a month how this could be something I could put my name behind it and ship it.

I was thinking this embedded software, it’s a little bit hokey and actually I’m not convinced we could get this into something I’d be able to stand behind within a month or so to ship it.

I didn’t realise until whoever it was put that on there, that this has stuck with me, along with many, many other things. What you’ll see or I’m going to talk about is some of the things that we want to do to sort out this middle bit.

Interestingly OK, this is another unrelated thing that seems suddenly to have of come related today or triggered by whoever chose that picture. You might have seen as well, we’ve been working with BBC and today they announced that there’s a new program that they’re gonna do around wearables. So although wearables are a fad and it’s not going to work, as we are God has told us.

Interestingly, this is about the BBC putting out some technology, so a lot of kids are they want to make a million pieces of this hardware can play with these wearable concepts. We might miss out on this generation, but obviously in the same way BBC micro in however many years, suddenly wearables are going to be interesting and real again.

One of many things over the last I guess 10 years that has led up to this. A bit of my background. I work at ARM. I’ve spent about 12 or 13 years at ARM. Initially started in R&D doing all sorts of transistor research, architecture research, compiler research, all this sort of stuff, building simulators. Which ultimately led to the v7/NEON which I was lead tech for. It was the architecture that kind of went into for ARM what was effectively the first smartphone chip design, if you like.

It was called Tiger at the time, it became Cortex, A8 when it went out there. This was a big bet for us and it kind of launched us from phones, where we were dealing with, into smartphones. For me, it was a very interesting project. It taught me how you take some technology, especially when you are working in R&D, you take some technology and then someone knocks on the door and sort of goes, umm could we make this real in a year? By the way, the project’s already started and you’ve got no choice.

It’s like, guess I have no choice then. Really interesting project and also a very insightful for me about how an industry, when it’s on to something, can afford the investments that perpetuates it. That’s really important. A lot of these things do come back to the commercial realities of being able to invest enough to allow them to move things forward.

You can start to see some of this stuff with IoT about how is the money going to be in place to make certain investments that you need to progress something. Then mbed is something that started quite a long time ago but has been thinking about I guess I saw ARM had it sorted as far as the smartphone.

There’s a lot of clever people at ARM, both technical and commercial, business. I could see that going, so I decided to dive down to this other end of micro-controller and see if I could add any insights there. So that’s my background, specifically at ARM.

Another thing, just to show you where my head is at in general. With Laura, who’s been helping with this event and Johnny who sat back there, we set up again, these things take time but we set up Makespace in Cambridge which is a 24/7 workshop you can go to. It’s got laser cutters, 3D printers, electronics, all this sort of stuff. You pay a monthly fee and you can go there. It might even be the biggest one in the UK, I don’t know. It’s pretty cool though. I think we are quite proud of what’s happened there.

From Audience: It’s the best.

Simon Ford, ARM: It’s the best, of course, yeah.

It doesn’t matter if it’s the biggest, OK. This is real, it’s very good. One of my internal objectives of doing this was initially, it started this was driven by stuff going on in my house and it outgrew my house. I wanted to see hardware/software, web and electronics all come together.

Actually, there’s a load of this social interaction serendipitous behaviour you want to stimulate. My personal goal for this was that in five years’ time after getting this set up, if we got it set up, a lot of IoT start-ups well, five was my goal, but I didn’t tell anyone. This was my internal goal. Five IoT start-ups could be traced back to being triggered or incubated or something in Makespace. I think we’ve already got one, in SimPrints maybe. For me this is really interesting. Doing these things, you learn a lot about Cortex thingy.

So, if anyone is interested, or you have friends who are interested in making things and bringing all of these things, this is alive and well in Cambridge. It’s totally self-sustaining. It’s a non-profit thing, but it’s £40 a month for 24 hour access.

Anyway, hopefully you’re starting to get a feel for where my head is at. OK. So that’s a bit about the background.

Now I just wanted to go through a bit of, you can look backwards to learn a lot about what’s going forwards. I want to set up why ARM has decided to, or perhaps invest in an area that is unfamiliar to it. Why we are willing to put our money where our mouth is and make this happen with a lot of our partners.

We tend to overestimate things in the short-term, but we tend to underestimate them in the long term.

I think probably everyone recognises this. The interesting thing for me as someone who is interested in developing technology is, by being part of ARM, we can actually take quite long bets.

Lots of start-ups can’t. They have to succeed or be acquired within a couple of years, especially in an emerging space. That means you have to, sets up the dynamics of the sort of investments you can make.

ARM is actually in quite a different position, in that we do actually have a very successful business with a lot of very successful partners. We have a long road map ahead of us of quite, I’d say, predictable success. Then you’re gonna laugh at me in three years’ time when something’s gone wrong, but it’s relatively predictable. That gives us a good position to invest in something where there’s a long-term payoff at the cost of short-term investment. That’s a nice sort of framing.

Just to look back, when electricity came into homes, it was things like motors that had a big change in quality of life. It meant our grandparents could count the number of electric motors they owned. Then the transistor came along this is obviously where it became interesting for ARM. Without electricity can’t do much anyway.

Silicon transistor had a massive impact and the analogous point is that our parents could count the number of things they owned that had a computer chip in it. ARM is very much around commoditising and sharing the cost of investment in a huge technology. Mobile technology has been the thing that’s allowed the industry to invest huge amounts of money in driving investment R&D, invention, innovation, capital investment in manufacturing and things like that to drive that curve. It’s the thing that made ARM as well.

What’s that meant is that fundamental technology can be reused in lots of different things that individually couldn’t have ever made that level of investment. That’s quite important to remember. The investment that has been by one industry can now be reused by lots of others.

Mobile is a massive industry. It’s quite singular.

At the moment, it’s quite specified what it’s doing. IoT is a little bit different to that. People are looking for killer apps, but there’s potentially going to be a very broad set of technologies and product. Each individually can’t necessarily make the level of investment that perhaps mobile phone has been able to do. So that’s worth remembering.

We’ve ridden Moore’s Law, or Moore’s Law has defined how much investment we’re going to make to actually get that transistor working at that size. Internet connectivity has done the same thing. The bandwidth is to scale as well. Ultimately we can still count the number of things in our house, in our lives that are connected to a network. It’s quite, it seems quite reasonable. You can probably count them in your head. You’ll forget a few, but it doesn’t seem implausible. The thing that we are aiming for is when this seems ridiculous. We succeed when suddenly we are teased about this by our kids, that it’s even a concept. That was an interesting thing.

I want to start from the position that we do have the technology to build from the silicon technology to ARM and its ecosystem have set us up for the hardware we need to build these things. While we have the technology, I don’t necessarily mean we have it today, but it is entirely predictable that we have things today, that we’ll have things in one year, we have things in three years. That’s entirely predictable that that technology is going to come to fruition.

It’s almost a non-issue. In terms of investment, it’s probably going to be reuse of technology that has got large markets elsewhere. So you’re sort of fudging it into slightly different application domains until you get big verticals where custom chips can be made for custom domains. There may be a bit of a lag there, but ultimately we have this underlying technology that perhaps 5, 10 years ago wasn’t in place to build these sorts of devices. It’s incredibly low cost now. It’s incredibly powerful and it’s incredibly available.

We’ve learned a lot as ARM about how that happened how that shared investment happened to make that happen. Just to give you an example, internally we did some experiments in some of the earlier days to try to understand this domain, given that we have the technology. This was a connective coffee machine that shows you what capsule are being used. We built this for a bit of fun, but also for a bit of learning. It’s not great, but it’s almost one removed from the internet fridge. People like coffee. It keeps them going.

The main thing we learned about this, it’s a very trivial application. You sense the colour of the coffee capsule. You have to calibrate it and you have an algorithm to work out. The idea is you obviously just send that back to a web service and it records this.

But you remember back to the BBC thing. Well, there’s a lot of stuff you have to do to supervise this stuff. You don’t have to worry about the actual technical details but you need to connect it. You need some technologies to make the connection out to the internet, in this case using constrained application protocol, like http, but for very low bandwidth connection over IPV6, things like that.

Now you need a server on this side that knows how to talk that. Web servers are very prevalent. Browsers know how to talk to them. Apps know how to talk to them. In this case, you are talking more about a device server that’s talking potentially a protocol that’s not http.

You need to make that connection persistent, because the device might go off the network, the connection may not be available. From a point of view of programming to that device, as a web service, as a web guy, you just want to not have to think about that. So you start putting persistency into your stack.

You’re talking on the Internet there, Hugo’s presentation about cost of investing in security and the need to invest in earlier is, I think, spot on. You need to secure your channel. Suddenly, this is looking a little bit complicated. You need authorisation. You need cryptography. It’s starting to get a little scary as well.

You then need to manage keys. You need to admin that. Well if you are securing the channel, you need keys to secure the channel. That means you need your device to be secure. You need keys, but you need secure storage for that and you need secure boot to guarantee the secure storage. So that’s getting quite complicated.

You need to bootstrap that onto the network to allow those keys to be inserted. Then you need to manage it, because if it’s out in the field, it’s definitely going to be wrong. You’re going to have to update that. You’ve got to manage that from the other side. You’ve got to load balance and to cluster it if you are selling more than one. You’ve got all these tools that actually allow you to build this stuff. Most companies won’t be building just one thing. They’ll be building lots of things. You actually want to reuse all these components. Suddenly, this simple application starts to look quite complicated. This really is trivial. I haven’t shown everything.

This is kind of what that first picture reminded me about. My feeling when we were doing that BBC thing was like I kind of fudged it and it kind of worked. But it was factors of, orders of magnitude away from being something that we would be legitimately shippable.

What this was leading up to is that IoT is a disruptive jump in the complexity of these embedded systems. As someone who’s wanted to find the reason disrupt the embedded development cycle, coz it’s very crusty and it’s the same thing that’s been around for a long time. This is really interesting to me. You now have a very real driver that is going to be this disruptive force. That’s one point.

There’s another two related points I kind of want to make to show the inevitability of how this is going to pan out.

Quote that has probably been overused. I read it in a slightly different way. I don’t care that the companies are less than three years old. The essence of this quote is that the groups, whether they are inside an existing company or it is actually a new company, I’ll probably go I’ve only started thinking about it around now. They don’t necessarily have legacy, because they are coming from a different domain.

They are not necessarily coming from the embedded domain. They are coming from a product domain, or a channel. They’ve got a distribution channel. They’ve got a route to market for a certain type of market. They’re not necessarily coming with any legacy in how they are going to build these things. That’s interesting because their expectations are different. Why does this take so long? Also, they’re choices are different because they don’t have legacy. They can acquire technology in very different ways than in a company that’s been developing iterative products over the last 10 years to get their market position.

The BLN is now Business of Software

Business of Software runs conferences for people that build great software businesses and products. To access talks online, hear about new events, contact speakers and stay in touch with great ideas, share your email address with us.


We'll NEVER sell your email. Unsubscribe anytime.

Suddenly you are going to get This is sort of are wearables are going to be made by fashion companies or technologies? A fashion company has no legacy but it can acquire technology in a very different way than a technology company trying to get into a different business.

Personally, I think this is going to be very disruptive in how the devices are built themselves. The way people think about the technology, what should be complicated and what should be easy is going to change. People’s expectations will be going to change and their situations that they are coming from will be very different. I guess that’s number two.

The last one, some people talked about this a little. When you go from just making an object even if it’s an electronic object to something that is connected and has a service and that service is persistent. Suddenly the supply chain looks very different and looks a lot more complicated.

This is a quite complicated supply chain, even in mobile, but it’s quite well understood. This is potentially a much more diverse market. This leads to a number of things, but one of them is obviously it’s going to be quite disruptive in how products are paid for and the business models that are used.

Fundamentally, these three things add up to this. How many people have read this book? Really? That’s not very many of you. You should go and read this book. There’s probably in fact, you should have a book club.

And still, people have not read the book? OK. So, Innovator’s Dilemma. I’m not going to tell you about it. You can go and read it. This one is more interesting, as well. Basically, for things to change sometimes, you don’t convince people, you just wait for them to die. This is my more cynical side. The mbed space is kind of like what it is, because the people who invented it haven’t died yet. This one is less politically correct, so this is generally the one I use.

I think you can all see a bit of truth in this. This is the suck-through-the-teeth. All that’s going to be hard. That’s what I’ve spent the last 10 years building my business to cope with that complexity for.

That was the set up for basically saying it’s inevitable that this space is going to happen. It’s also inevitable that there’s going to be quite a lot of disruption. One thing we can do is learn from history.

I was going to talk now a bit about a business unit we set up in ARM and why we’ve done it and what we are setting out to do. It’s that background that has really given us the confidence to go and make this sort of investment.

If no one was clear, this is where you are right now. I think it was mentioned before slightly. You are right up here, ready to come down this wonderful ride up this hockey stick off here.

We as ARM in the embed unit we set up is aiming out here. The important thing I wanted to highlight is this: 5 to 10 years. We are not setting ourselves a two-year start-up goal. We are sending ourselves a 5 to 10 year goal. We think that’s necessary to work at those sort of time scales to actually deliver the sort of things we are doing. There are some nice benefits of this. This little right down here, also includes lots of interest start-ups with clever people that don’t make it. We are quite happy when they join our team. This is where we are aiming.

Oh, yes. We are hiring lots and lots of people in lots of it very interesting areas. So if you know people who are looking for jobs in this space, especially techie, web or embedded, those sort of guys, please tell me afterwards.

This is where we’re aiming. We’ve seen this sort of thing before. We know what makes these sort of ecosystems successful. For mobile, we invested obviously, a lot in the process of technology that was going to be used in mobile. It was also investing in things like the standards, the Internet standards that we’re going to need to be running on these mobile devices, the sort of ecosystems that made it possible and the sort of platforms and development changes that kind of were the definition of moving from phones, where it was a product manufacturer making a phone to smartphones, which was much more beyond the device itself and beyond the device itself in terms of development and all of that. We obviously have been very intimately involved in this. We think there is a huge amount to learn from this.

In some sense, IoT looks like the app part of the smartphone market. That gives you some sense of this sort of diversity that you don’t necessarily get the same consolidating effects that the phone does. But you actually want more of the behaviour of the app ecosystem that was built around the phones. It just happens that app might be a physical thing.

People sometimes forget the I in the IoT, but there is a lot of work that needs to be done investing these sort of standards. We are working with a lot of these standards bodies to make sure that these standards not only get written, but that they are actually available in implementations. A standard is only really valuable if they are good implementations. That’s why I think Ethernet was so successful, right. There were good implementations, not just a nice pretty well-written document. We’re making a lot of investments in these sort of things.

The sort of things that are inspiring us, something like Android’s full stack, but very domain specific operating system. It’s very clearly for a mobile phone. When RAILS came out, it was all about baked in testing and conventions that allowed people coming to this industry. If they were engineers before, maybe they moved to web. This is actually how you build a web application; this is how you do testing. Things like Node for inventing frameworks. NPM for component management. These are our inspiration. It’s really about just catching up on the device side with the rest of this world.

So, what did we do?

Well, at Techcon last year ARM has a yearly developer conference in California. So, end of last year, we basically said, we’re up for doing something if you are. This is not something we want to do alone. This is something we want to consolidate investment through a lot of different companies. We’re not trying to build something ourselves and then sell it. This is about investment getting the level of investment we think is necessary to move this industry forwards. So we did a bit of the raz-ma-taz, it’s the show. We said, Hey! This is what we’re thinking. We didn’t quite phrase it like that, but this is what we’re doing. The response was really, really positive, from lots of our partners. That really signed the deal where we’re gonna invest in this.

The BLN is now Business of Software

Business of Software runs conferences for people that build great software businesses and products. To access talks online, hear about new events, contact speakers and stay in touch with great ideas, share your email address with us.


We'll NEVER sell your email. Unsubscribe anytime.

Effectively, what we’re doing is we are investing in an operating system for these sorts of devices, or the tools you need for these types of devices and building an ecosystem of those sorts of people that can incrementally benefit from adding their piece of the puzzle to this investment. This proper investment, not only in technology, but in ecosystem building to allow that consolidated effort by a huge number of companies into something that moves the industry forward.

It’s not something that a company can do alone. As you’ve seen from the earlier things, it’s not something that a company could probably afford to do. This is a concerted investment by a lot of different people into something that everyone can share.

I’m not going to go into too much of the details. It’s more about the way that we’re approaching it. We’re building operating system. We’re building the sort of server technology that allows to get the data in these devices into these web stacks. Fortunately, we’re investing a lot in building that ecosystem. The working relationships that allow a project like this to be successful. That in itself, the social engineering aspect is incredibly hard as well.

There’s some technology. I’m not going to go into the details, but I just wanted to show you that this about a consolidation of technology as well. And to give you an example, we are starting to acquire companies where these technologies reside, then making that more widely available. The most recent acquisition that we announced was PolarSSL. This is one of the world’s leading security TLS/DTLS implementations. We’ve acquired that and we’re going to make that open source under a permitted license. That technology is now available to people. These are the sorts of investment we are going to be making with the rest of our partners to allow this step forward.

Just one other thing I just wanted to mention, because it seemed to be a theme of some of these talks. Security is something that a lot of people are scared of and that a lot of people get wrong. A lot of what we are doing is baking in the security aspects of how you manage a device, how you secure a device, how you boot a device, all those sort of things into this space, operating system. There’s even some social engineering going on about how people use these security aspects so that they don’t make mistakes.

It’s not just technical. It’s actually providing the approaches to doing security on the devices so that we have a better guarantee. As Hugo said, this is not something that each company can afford to do. Everyone has to invest in something to really make this happen.

The interesting thing for us is effectively this sort of hinge point we can create with companies that would worry about the cloud and companies that would worry about the devices. That was what we are trying to put together. With this level of disconcerted investment in technology from a lot of companies, this will really help move this industry forward and make the device side possible.

I think that’s my time. That’s all I wanted to talk about. Thank you very much. I hope that was interesting.

The BLN is now Business of Software

Business of Software runs conferences for people that build great software businesses and products. To access talks online, hear about new events, contact speakers and stay in touch with great ideas, share your email address with us.


We'll NEVER sell your email. Unsubscribe anytime.