 I'm Matthew Locke, I'm the Director of the IoT and Invented Group for Lunaro. So this is Light? Light, for short. And there's lots of companies, it is a very active area at Lunaro. Very active, we have 10 members currently including NXP, who's booth we're at today. And everybody's in a rush to get IoT to work with security and everything? Yes, one of the biggest problems in connecting all these devices to the cloud or the internet is making sure that they're secure so no one can hack them and get your personal data or cause problems with a device. And so one of the platforms that I'm being worked on right now is Zephyr, right? And maybe you can introduce? Zephyr is a real-time operating system hosted by the Linux Foundation of which Lunaro is a member as well as NXP. And Maureen here is one of the maintainers of the ARM subsystem of Zephyr. Hello, so who are you? Hi, my name is Maureen Helm, I'm a software developer at NXP. And you are the maintainer of the Zephyr area? Yes, one of the maintainers working on the Zephyr project, focusing on support for NXP devices. So there's a lot of activity with Zephyr right now? Yeah, absolutely. We've seen a lot of activity in the community adding support for new SoCs, new boards, new devices. And what is special about Zephyr? Is that the right way to do security? So I think what's special about Zephyr is the open source development model. I think the fact that anybody in the community can contribute code and take a look at patches, review patches, comment on them. I think that's something that's really special about Zephyr. Is it an RTOS? It is an RTOS, yes. And there's lots of RTOS in the world. How do people choose not to? Oh, how do people choose? So Zephyr is a flat memory model RTOS and one of the nice things about it, in addition to the community aspect, is that there's no legacy implementation of security or some of these other important features to make sure it's secure. So we are starting from a blank slate designing for the next generation of devices that are connecting to the internet. So it's very, very exciting. So working for the next gen. So right now it's running, here you have a demo on a Cortex M4 but you're talking about the Cortex M33 or is that what it's working for? Well, what I mean by next generation of devices is the actual end products that people are developing which are going to be the Cortex M4s today. That's going to last for a while and then ARM has new cores coming out that make it even easier in the hardware to make sure that these devices are secure. So it's been a year or so, a little bit more that Leonardo's involved in the lights, right? Leonardo's only been involved for what we had like eight months. Zephyr project's been around for over a year now. NXP was one of the founding members. Many of the Leonardo engineers, they volunteered to come over from what they were doing in Cortex A and V8 to do V8 M stuff. Exactly, a lot of the developers are coming from the Linux side of things and working on Cortex A and being very familiar with working upstream in the Linux kernel and we're bringing that to the RTOS world in Zephyr. So I was working on general software enablement for Kennedys and LPC devices and so kind of creating a common enablement strategy so that means sims to support peripheral drivers and things like that. And how do you choose what gets in and what doesn't get in? Because that's your job, right? What gets in the Zephyr and what doesn't get in the Zephyr? It's a combination of things. We're definitely doing a lot of code reviews and so it's pretty rigorous reviews from a maintenance model. You have to look at all the code? Yeah, we have to look at all the code and so we have a lot of people in the community that are looking at it as well and so we have experts from the various arm vendors not just in XP but within the Lunar group that are also looking at it. Is there a lot of code or is it not? It's supposed to be very complex as a whole point, right? So is it as small as possible? It is as small as possible. However, to get these advanced features like connectivity and security and something small is actually very hard. So the code ends up being, I would say, almost more complicated in some senses and as well as the volume of code there's tons of features to put in there so configurable it still stays very small but you have the options to choose these different features. So it's like crazy advanced algorithms you need to put in there to make things work? No, I don't think I'd say that. I think that there's a lot of configurability in the sense of what SoCs are working on so you're not going to use everything for that particular device, right? You have an XP device and there's a subset of what we use in the Zephyr kernel that runs on our device but if you switch to another vendor's device that configurability aspect of it really helps. So are you able to run Zephyr on this little thing? What is this? Can you hold this? Oh, this actually is not for Zephyr. But it says BLE Heart Rate Monitor Zephyr? Yes. Is this one of those? No, this actually goes over here, so I'm starting on that. And this is the MCU stuff? Yeah, so Zephyr is running on this device. This is a Kinetis K64F, which is a Cortex-M4. That's running the Zephyr Bluetooth post stack with a peripheral heart rate sensor profile. It's one of the out-of-the-box sample applications that you can download the Zephyr so you can get started with this today if you're not doing any kind of magic for the demo. What's the Zephyr development kit from NXP? Well, NXPware was a relatively new development kit that was actually with a partner called Micro Electronica. But it fits really well into what Zephyr is trying to take in the fact that there's support for a lot of different sensors built into the device. So we have accelerometers, gyroscopes, so we're talking about Internet of Things and sensors support. It makes it for a really great development platform. All right, so it's going to work out. The Zephyr is going to be great. And in our light, it's going to work out. Absolutely, yeah. There's a ton of work to do. We're going to be working for quite a while on this project. Cool, thanks a lot. Thank you. You said there was 10 members. So there's NXP. What kind of other members do you have in the light group? ST is one. Right there. So they also have stuff going on with that. There's no demo. So the embedded world is what you're working in. Embedded world is what we work in. Our goal is to bring the embedded into this next generation. That's what IoT is always about. The embedded's been around for a while, and IoT is really taking the embedded world and connecting it to the cloud. There's a Cortex M23, M33. Is it going to bring you even more stuff to do? Yes, there's some very advanced features, especially related to how you manage memory so you can isolate tasks from each other. What we're working on is an idea of bringing the container concept from Linux to the microcontrollers. We're calling it microcontainers. You want to be able to isolate your connectivity stack from the cloud. So if you have a cloud application that downloads some algorithm into your microcontroller to do something, or even a Cortex-A device, then you don't want, if there's a bug in that, you don't want it affecting into your RF configurations and your IR stack. You want to keep them completely isolated from each other and only allow them to talk when they've been allowed to. And so there needs to be security in those microcontainers. Yes. It's not RPM or something like that, small packages. And people are doing it in different ways. Ubuntu Core is doing it in one way with the Snapcraft, and they're also a member, right? Ubuntu is also a member. A big thing what we're doing in Light is kind of end-to-end. So you can't just talk about the device, you also have to do the gateway. And the gateway is going to be running distros like Red Hat and Canonical, who are members of the IoT embedded group at Linaro. Because they need to talk the same language, they need to be able to securely talk to each other. The devices need to be able to authenticate to the gateway, which then authenticates to the cloud. They need to be able to send data back and forth. You may have intermittent connections between the devices that are not always on and the gateway. So they need to be able to cache data and then send it up into the cloud. And the opti is a big part of this. Opti on the gateway is going to be a big part of this story, definitely. We're going to have containers on the gateway. We're on the Cortex-A Linux stuff. We're going to have microcontainers on the RTOS and the devices. And, of course, you've got the whole cloud container. So it's this kind of end-to-end idea of isolating tasks from each other to ensure the security of the device. It sounds really awesome. It's pretty cool stuff. It sounds like cool stuff. Even though people have been working in a better world for decades and decades, this is groundbreaking. This is the front of what the future of embedded is going to be. Exactly. That's really what's exciting and new about this. As you said, the embedded world has been around for a long, long time. Doing stuff on microcontrollers is not new. Doing stuff with Linux on these kind of embedded devices is not new. What's new is tying them all together like we just talked about. Well, it's never really been connected to the internet or to a cloud. It's doing kind of isolated tasks that has no connectivity. It's really bringing that connectivity piece that's changing the whole market. I guess ARM is counting a lot on you to get this right because they need your stuff when the chips are ready for the M33 and M23. It's going to be a big part of those solutions. That's really the point of Linaro, is making sure that the software is ready for the ARM ecosystem in general as new technology comes out. That's exactly what we're doing. At this show, I saw the first chip at the ARM booth. It was an ARM M23. The 23, right. These guys are in a race right now to get the 33, 23 out. There's going to be a bunch of solutions coming out. They're going to have a trust zone on the Cortex M. Exactly. And I think we'll see a lot of that in next years in embedded worlds. It's going to be very different than this year. There's going to be an uptake in the open source security systems. It will be very much counting on the OPTI to be a perfect solution. That's right. You can see the start of myOS is here with the booth. And there's a few other open source R-tosses that are here. You can see in the uptake.