By Patrick Mannion, Contributing Editor
Prior to becoming vice president of software at Silicon Labs, Skip Ashton was VP of engineering and technology at Ember before Silicon Labs bought it in 2012. Over the years, he has been listening intently to customer needs with regard to low-power, low-cost connectivity while also tracking and contributing to standards efforts, including zigbee, Fairhair, and the Connected Lighting Alliance.
Q: Skip, what can you tell us about some of the more exciting things happening in IoT?
A: For us, the most exciting thing is just the tremendous growth that’s going on. We’re seeing 40% to 50% year-on-year growth over the past few years, and it looks to be accelerating, not slowing down. Connected devices — be it over Bluetooth, over 15.4, and even our proprietary stuff — is just an exploding area. Mostly what we’ve been focusing on are the low-power connected devices and their usage in the home, small-business, and commercial buildings applications.
Q: Do you see any misperceptions about IoT out there?
A: To a certain extent, we’re in the trough of disillusionment. There’s a lot of security fear and concern, and there are plenty of reasons for that, given some of the poor security implementations out there. There are also concerns about interoperability and how to unify these protocols a bit so that we that don’t have different towers of things that don’t talk to each other.
Q: What’re some of the biggest challenges for IoT?
A: For device makers, it’s about how to build a product such that it’s still going to be relevant three years from now and I don’t have to keep going through the churn cycle of what protocols and what application objects are defined.
Q: What do you see as the “killer connected device?”
A: Beyond voice control, which seems to be a killer app waiting for things to attach to it, it’s very personal and use-case-specific. There isn’t a great unifying theme. Users may decide that they want a connected device because they really love Alexa or Google Home. Maybe it’s a smart lock, or maybe it’s a light bulb. But everybody’s getting into it for a different reason. It’s hard to identify the thing that you have to have in your home that, when you’re using your Alexa or Google Home or Siri, you’ve got to connect this device to it.
Lighting is probably seeing some of the best growth, but does that mean that a connected door lock is a good idea for me? It might, but it might not. Also, if my lighting guys pick protocol A, and my door lock has protocol B, how do they actually interoperate? How do those industries align? (See “Balancing interoperability with efficiency for wearables”)
Q: There’s a lot of interest in industrial IoT: Could that be the “killer app?”
A: I think it’s safe to say that there are a lot more units in the home. Industrial tends to move slower, but in this situation, industrial gets higher price points and they tend to care more about the cloud connectivity and remote monitoring. A home-user just wants to do a smart lock or something. Monitoring, train engines, and airplane engines have gotten amazingly good traction because there are these phenomenal business reasons around it. But they’re a different volume-use case than, say, a light bulb. There are a lot more light bulbs in the world.
Q: With predictive maintenance, you can use data analysis so that you can save millions of dollars by not having a pump go down unexpectedly.
A: This will sound funny, but there’s a rational business process and decision-making process that comes with monitoring pumps and doing predictive maintenance. Homeowners’ decision-making processes are anything but rational. It’s not a business decision, it’s more about wanting something that’s cool and makes them feel good because it’s electric and they can show it off to their friends. So it’s a very different set of drivers. And we’re much more focused on that sort of home and consumer and even commercial building use case, which I think is very different than in industrial.
Q: What about wearables and health monitoring?
A: The consumer driving the wearables is much different than medical driving the wearables. If your doctor says to wear something, you’re going to wear it. Whereas if you want to buy a Fitbit, you’re doing it because you think it’s cool, but you may stop wearing it a month later.
Q: What other challenges are facing IoT developers?
A: Along with interoperability, security is one. We’ve gotten to the point where everybody understands the need to be able to do over-the-air updates to their devices to improve their capabilities and their security, but it’s not all the way there yet.
Q: So what would you expect at least on a baseline device to consider it somewhat secure?
A: We look at security at multiple levels: device security, the protocol security, the application-layer security, and the ability to update and reflash a device from my cloud-based service or from my phone app. With the device, make sure the debug ports are locked. Also, protocols have come a long way. Look at Thread as a more recent generation. It has much more robust and heavyweight security built in than stuff that came out five or 10 years ago just because, A, we can count on faster processors and better security accelerators, and B, there are some better security protocols that we’ve now been able to fit on these small devices to do that. So the protocols are more secure.
I view security kind of like water: It runs downhill to the simplest point. Still, unfortunately, it’s easy to find devices that are simple to hack into, but you’re starting to see more devices that have what I’d call a base level of security that’s reasonably respectable.
Q: Are there security enhancements coming our way?
A: The next generation of devices will go to a full-secure element on the device. So sensitive material is held within a secure element, and that means that I can now fully encrypt my software downloads and things like that.
Q: What about subsystem and functional partitioning and protected memory?
A: When you get down to the small devices that are 256K of flash or 512K of flash, we see much less of a demand for device partitioning. There really isn’t a concept, today at least, in which I want to download a third-party app onto my light bulb or door lock. That kind of doesn’t make sense on these kinds of devices. For a smartphone or some of these larger Linux-based gateways, you’re expecting to download third-party apps, so it’s important to partition that.
Q: Is security in conflict with interoperability?
A: Yes, but the only really good way to have both security and interoperability is to use well-known protocols. By definition, if I make up my own security, A, it’s not going to be interoperable with anybody else, and B, I’m probably going to mess it up. I mean, let’s be honest, we find security vulnerabilities in these well-known protocols that have been around for a decade. So if someone is making one up, the chance of it being as secure, as scrutinized, as analyzed as something that’s been around for a while is kind of unlikely.
So there’s a reason we don’t do new and novel things in security that often. Security guys tend to be very frightened of new and novel. Because usually that means hackable and it’s got some other flaws in it, and it’s just a matter of time for them to come out.
The benefit of everyone on the planet being seemingly addicted to a smartphone is that we’re using the smartphone as a configuration and commissioning tool, which means having to put in a passphrase or some short set of digits to commission my light bulb onto my network. This isn’t a very high bar, but it’s nearly impossible to commission something like a smart plug any other way because it has no GUI. So if I can assume that a smartphone is in the ecosystem somewhere, or that you have one, then all of this security gets easier, and all of this configuration and setup gets easier doing it in a secure way because I can assume a user interface.
Q: I like Electric Imp’s approach, in which they use the phone to send an optical signal to the sensor. Any other novel approaches you’ve seen?
A: Well, the other interesting one is that with Amazon, those dash buttons are, I think, doing it acoustically. I don’t even have to have a user interface: My button can talk to my Alexa through things that you and I might not understand but that it understands.
Q: So I guess it really kind of hits the road when you’re resource-constrained. How are you addressing such applications?
A: And this is the age-old dilemma. Things have gotten better, faster, and stronger over the years thanks to Moore’s Law and lots of engineering. So many ask, “Why worry about code space and other resources when they’re getting better, faster, and stronger?” We actually follow Moore’s Law down, not up, because our customers want the same device at half the cost, not double the device for the same price. There isn’t a price for that door-window sensor that’s too low.
Q: What has Silicon Labs done to improve the design of IoT devices from a product point of view?
A: We have an app that you can download from the Google Play store that does Bluetooth commissioning of zigbee devices, for example. So the device comes up in Bluetooth, we commission it, then it flips over to zigbee. And we have customers using that today.
Q: So it’s multi-protocol support. Does it need two radios?
A: On the low-power side, we have a lot of people using zigbee. Thread is a growing area and we do a lot of Bluetooth, but we also do a lot of proprietary. So it’s a multi-protocol chip so I don’t have to have two of them. It provides a nice, seamless user experience, and so it doesn’t add a lot of cost for the manufacturer.
Q: What do you want to see happen in terms of enabling that next generation of IoT? What advice would you give designers?
A: The biggest advice that we give people now is to future-proof designs because the industry is in such a state of flux. Design it to do over-the-air updates to change protocols and its behavior over time. Predicting now how your device is going to be used in a year — never mind three years or five years — is extremely difficult.
Q: Are there any things that you see that are obstacles that you want to say to designers, “Please, somebody fix this before we go crazy?”
A: The biggest fix for us that holds people back that we hear is ease of use — letting a consumer take a device out of the box, install it, configure it, connect it to whatever it needs to connect to seamlessly to give them that user experience without them banging their heads against the wall in frustration. This is really a huge impediment. There’s a reasonably large percentage of devices that ends up back at the store or gets returned online, and they’ve really never been used because the setup and configuration experience was painful enough as to make it useless for most people.
Q: I’m in this industry and I recently set up a different set of connected light bulbs. And it took me two-and-a-half hours to get the gateway set up and configured. Do you think that an average user is going to spend two-and-a-half hours to get a gateway set up to connect five light bulbs to it?
A: So you may make the best light bulb in the world, with the finest color control and that sort of stuff. But if your smartphone app and your gateway connectivity are slow and painful, it doesn’t matter because I never had the chance to check out your light bulb.
Advertisement
Learn more about Silicon Labs
