Our guest today is Ross Sabolcik, the Senior Vice President and General Manager of Industrial and Commercial IoT Products at Silicon Labs. Silicon Labs is a renowned technology company specializing in the design and sale of semiconductors and software for manufacturers in the Internet of Things infrastructure sector. Ross boasts an impressive track record with over two decades of experience in this field and holds a Master's Degree in Computer and System Engineering, as well as BSEE from Penn State University.

In this talk, we will delve into the advancements in edge connectivity for industrial IoT devices. This will include recent breakthroughs in cross-technology compatibility and simplified programming. Additionally, we will explore the underlying value proposition of three recent and upcoming product launches, namely FG-28, GA, and Series 3. Our primary focus will be on how these innovations facilitate machine learning capabilities at the edge.

Don’t miss out! Tune in to learn more.

Key Discussion Points:

• What innovations does Silicon Labs introduce to the market, and what are the prevailing industry trends?
• In what ways does Bluetooth technology change the cost of connectivity?
• The importance of security for these edge devices

If you're curious to know more about our guest, you can find him on: 

Website: https://www.silabs.com/

LinkedIn: https://www.linkedin.com/in/rosssabolcik/

Erik: Ross, thank you so much for joining us on the podcast today.

Ross: Hey, Erik. Thank you. I'm excited to be here.

Erik: Yeah, I'm looking forward to this one as well. This is going to be a conversation where we deep dive a couple of your newer technologies rather than trying to cover the entire business. Really looking forward to hearing what's coming out in the wireless space. But before we get into that, for the sake of our listeners, let's start with, let's say, a three- to five-minute overview of where the business is coming from, with a particular focus on industrial.

Ross: Sure.

Erik: So who is Silicon Labs, and what do you bring into market?

Ross: Silicon Labs, we're a fabulous semiconductor company. We're really focused exclusively on the Internet of Things. About 10 years ago, we started our journey on the IoT, even before it was called the IoT, building up a collection of wireless products that would allow customers to easily connect their devices to the Cloud. Of course, it wasn't called the Cloud at that time. But if you look at what we've been able to do over the last decade, it's really build up a comprehensive wireless portfolio of SOCs that target embedded developers and help them with that wireless journey.

We started off with Sub-GHz protocols. Those are used in applications, if you think of garage door openers, or home security systems, or even things like a smart meter, those have used that Sub-GHz technology because that's extremely good range and reliability. In addition to that, we are the market leaders in 15.4. That is the wireless technology that underpins standards like ZigBee or Thread, which is the basis now of Matter that you're probably hearing a lot about. We've also added Bluetooth capabilities into our portfolio. Most recently, we've added Wi-Fi. So the Internet of Things is not one wireless technology for every application. It's a collection of a whole bunch of different wireless technologies that are purpose-built for given applications.

Our goal as a company is to offer customers that full portfolio of products, that regardless of what application they need and what type of data rates or range and reliability, we have the right solution for them. We also focus a lot on compute capabilities, so you need to have a certain amount of intelligence at the Edge. That intelligence has continued to increase over time. Our products include very capable MC use and peripherals and accelerators that are built into them to help customers not only be able to send wireless data but to do the analysis and pre-processing they need to do before they send it. We are 100% focused on the IoT. It's really all we do. Our goal is to cover all of those types of applications, regardless of what the end customer is trying to build.

Erik: Great. I look forward to getting into this topic of edge computing. I think there's a lot of growing demand there, but let's cover that in a moment — the application areas. So we're going to be more focused on industrial IoT. But I know you're active in smart home and smart city and retail and connected health. Is there some kind of 80-20 rule where the company tends to focus, or is it highly distributed across these different markets?

Ross: As a company, we're broken up into two business units. I oversee our industrial and commercial business unit. We have another business unit. That's home and life. If you look, we can say, roughly, it's about 50-50 in terms of our revenue between those two business units.

In industrial and commercial, we're organized into segments. We have three main segments we focus on. We have Smart Cities. This is things like smart metering networks, street lighting, kind of metro area applications, these long range, ultra-long-distance applications. We also have a commercial segment. That's focused on things like retail. Think electronic shelf labels or point of sale terminals. We also have in their commercial gateways in clinical, medical, and then industrial segment. Then that industrial segment, its applications, you would imagine that would be their — think predictive maintenance, true industrial automation and control or things like that. Then on the home and life side, we're a very major player in the smart home. So we have a lot of applications there, be it door locks, or light bulbs, or thermostats, things like that. Then we have a Consumer Life Division, which is focused on consumer health applications.

We cover a broad range of applications. I think that's really important because we're able to invest in certain areas that are major drivers for given technologies, and then leverage that technology in those other areas where if we were only trying to invest in one or the other, we probably wouldn't be able to bring as much R&D dollars to bear on certain technologies that are more prevalent in one segment versus the other.

Erik: Yeah, it makes sense. Okay. Great. Well, we're focused on the IIoT space today. I think there's a lot of progress that's been made there to some extent. Forced to be made over the past few years simply because COVID made human operations a lot more challenging than they were historically. Therefore, there was a lot of incentive to automate things with IoT solutions. And so I imagine that there were some use cases that that really pushed forward. Now we have a separate set of challenges, at least in the US, which is around bringing manufacturing back and then figuring out how do we do that with a very high cost, scarce labor source, which again drives adoption for a lot of use cases.

So I'm curious. If we look at those two, if we look at COVID which we're coming out of but has already pushed a lot of companies that we're playing with POCs now to actually adopt things at scale, and then we have this situation of figuring out economics for manufacturing in higher cost locations, what are the big trends that are moving markets for you over the past couple years?

Ross: Yeah, I think that's a really great way to frame it. I think supply chain, as we're looking at COVID, was a huge motivator for a lot of investment. If you can go all the way early in the supply chain which is tied to asset tracking — where is my whatever, my stuff? Where's my widget? How is it progressing on its journey? What are the conditions that's experienced on the journey, if you think of cold chain monitoring — that early-stage supply chain, I think, drove a lot of what we saw during COVID. People really started to appreciate how important the supply chain is and how important the logistics of moving things from place to place really are. We see a lot of activity that's come out of that in supply chain. Be it smart pallets, pallets that actually have wireless connectivity, so you can understand where they are again in that journey. That can be all the way down to tags that would be on products or boxes that would be on that pallet, so you could see where exactly is my stuff. So that's kind of in the supply chain, we saw a lot.

If you play that forward to even on the consumer side, on retail, we saw a lot of interests in, how do I automate my operations, how do I make myself more efficient, how do I take advantage of the scarce labor that I have? And so electronic shelf label is something we're seeing also on the retail side that got a big boost during the pandemic. The real, to me, illustrative anecdote of that: I went into my home supply store. You probably remember that lumber have wild swings on the pricing of lumber. And so my local home improvement store, they put electronic shelf labels on their lumber, because you can change prices dynamically. If your supplier is telling you, "Hey, your lumber just went up 50%," you can go ahead and change that on the fly and reflect that in the cost of what you're trying to sell. Across that whole journey on supply chain, we saw a lot of those applications really accelerate during the pandemic. I think what we're seeing now is, you're seeing a re-engineering of those supply chains, sort of that onshoring, and also the need for a lot of flexibility in the supply chains.

From manufacturers, we're seeing an acceleration of the adoption of wireless technology in the true factory floor. Factories are an interesting use case in that they're highly automated. They're collecting lots of data already across their operations, but it's tended to all be wired. Wires are very reliable. It's relatively inexpensive after you install the wires, but it's not very flexible. So if you need to reconfigure machines or cells, having to redo a bunch of wiring is not as attractive as if you have wireless, and you can reconfigure those cells. So we're seeing a big push now what we would call the wired-to-wireless transition in a lot of that industrial space. That gives you that flexibility, that if you need to change what you're manufacturing, your cost of reconfiguring go down significantly. That's an area we're seeing a lot in the true IIoT.

The other trend that we're seeing there is really tied to predictive maintenance. So now that it's become very inexpensive to add wireless to any type of sensor, we're seeing a lot of retrofitting of machines to try to look for early signatures that the machine may need maintenance or the machine is about to fail. Since a lot of these are retrofits on existing machines, wireless is really attractive to do that. Because you don't have to go and pull a bunch of new wires all over your manufacturing site to start to collect that data. So predictive maintenance is another area we're seeing really starting to accelerate.

Then the third thing, I would say, is in that HMI, that human-machine interface. We use a term that Bluetooth is the new RS 232. Historically, a machine would have a hard-wired port you could plug into if you wanted to do diagnostics. You'd have to have a dedicated box to get those diagnostics and understand what's happening in the machine. With Bluetooth as an interface, you can have a technician that has their cell phone or a tablet with an app on it, and now you can connect to that manufacturing machine and understand what's going on inside. So those are some of the big areas that we really see a lot of, I would say, innovation happening right now in the IIoT.

Erik: Over the past few years, we've had steady improvement on the technology. We've also had rapidly increasing volumes in at least some of these categories. How has that impacted the price point for some of these solutions? You mentioned that it's now much more affordable to connect brownfield equipment, for example. Can you share a little bit about what the cost trend looks like and where we are if we wanted to, for example, connect the sensor to a machine, what might the cost of that connectivity look like today?

Ross: Yeah, I think particularly when you're talking about standards-based connectivity, there's really been — that's the nature of the beast in the standards. They're very highly optimized in terms of cost. So if you're looking at, let's say, adding Bluetooth and there's a component of what it costs to add our ICs. We're a semiconductor manufacturer, so we're one part of that system. But you can start to talk about like with Bluetooth, sub-dollar kind of price point, to add that connectivity into a module or a control box that you would be looking to automate. From the wireless connectivity side, it's gotten really inexpensive. There's also, normally tied to that, there's some amount of compute. What we see happening on the compute front is more capabilities, more accelerators, more memories, more peripherals that you could get with that wireless technology you're putting in where you could run the entire system on a single chip.

That example I gave to you of electronic shelf labels, there's really just one IC, one semiconductor device in that entire system that is doing the wireless connect communications, running all the software that's required in that application. So if you think of a tag, those are highly optimized. As we move forward in time, I think you'll see even more sophisticated type of end node devices have the wireless and the peripherals and the compute all integrated together, which really helps drive costs down even more.

Erik: Great. Thanks. Well, let's get into some of the products that you're bringing out that are related to these trends here. Maybe you can start by giving us just a high-level preview, and then we can deep dive them one by one.

Ross: Sure. When I talked about those segments that we cover in this industrial commercial side, we start off talking about smart cities and smart metering. Those are applications that require really long range, super high reliability. Historically, they've been done with Sub-GHz radio technology — radios that are running in these ISM bands that are under one gig. We have a lot of products in that space. We also talked a little bit about Bluetooth for that HMI, human-machine interface. So we're about to introduce a new product that will meld that Sub-GHz technology with Bluetooth in a single device. So you can have that range and reliability if you're connecting devices over these vast distances, but still then have local connectivity if you're going to do HMI so that technician can go up and diagnose without having to connect the cable or without having to add the cost of another radio. That's a new device we're about to launch. We call that our FG28 that has that Sub-GHz and Bluetooth together. We also see those use cases in industrial for Sub-GHz. There's some wire replacement type of technology that's looking at Sub-GHz because of that range and reliability that will also be beneficial there.

The other big announcement we have is around our Series 3 platform. So if you look at the history of our roadmap, we've had three or two major generations now of these wireless SOCs. We had our Series 1 which is where we first put together the wireless and the compute into a single device. That's been a very successful platform with us. We're currently shipping our Series 2 where we've added additional types of connectivity, I think Bluetooth and Wi-Fi on that platform, with even more compute. Then Series 3 is our platform that we're building where we're really amping up intelligence at the Edge. The ability to do more processing, the ability to have accelerators around things like AI, ML, to get more processing capability, higher levels of security, better kind of software development experience, that's a whole new chassis that we've just announced that are works with conference that we think is really important for taking IoT the next step on its journey.

Erik: Great. Well, let's take them one by one. The FG28, it sounds like the big value proposition there. At least one of them is reducing the amount of labor that would be involved in maintaining devices. I think that's always a big part of the cost structure. You say, okay, I've got this device. It costs $100. But then, every time I want to have a person touch it, that's going to cost me $50 of labor of sending somebody out there and doing this. That can really kill a business case. So where are we today in terms of being able to deploy a device, and then ideally never have a hand touch it again? I mean, what's the range? What are the situations where that's feasible and not feasible? How does this new product impact those situations?

Ross: We talked about smart metering as one of the real poster child applications for this. Historically, in that smart metering space, the first smart meters were all about: I don't want to have to have someone go walk by your house to read the numbers on a meter to bill you. Those first wireless systems were really driven by, how do I remove that labor cost? The first order now, that labor cost has gone away. Those early systems were these AMR systems, these automatic meter readings. One of the things I see across all these applications is, once you wire up this network and you have a killer use case that pays for it — billing was kind of the killer use case in metering — customers start to ask, what else can I do with it? I have all this wireless connectivity. How else can I leverage it? In that metering space, what we're seeing is, utilities are wanting to use the meters as smart sensors to understand what's happening in the grid. That's a trend we're seeing with the new meters. Great. Buildings are given. But now I want to get Edge intelligence. Having that wireless connectivity allows you to get a much better feel on what's happening across the grid than you could if you didn't have these, for example, hundreds of thousands or millions of meters that were monitoring the grid.

If you think about now, instead of just saying, "Well, I'm going to build this next," you can start to look at as the power out at their house. Am I starting to see brownouts on my network? All these types of metrics that you can monitor to get health. And so that's been a big change. To your point, now you can get those metrics from the smart sensor network that you built instead of having to either have a dedicated network just doing sensor or having to put a technician out to take those types of measurements. We see, as the wireless technology or the connections increase, now you're doing more and more monitoring of more and more things, reducing that need to actually have human intervention. So we see that a lot as these networks start to build out.

The other trend that we see happening is, there's been really a move to standards-based as opposed to proprietary networks. So a lot of factory automation, a lot of these smart grid utilities, historically, every vendor had their own solution. We see now move to end customers don't want to be locked into a single vendor, so they're asking for kind of open standards. In metering, that's like Wi-SUN. You'll see, in the industrial space, you'll see other ones. So WirelessHART, as an example, is an open-standard based on wireless technologies. So that allows you to not only deploy these systems and take advantage of all the data but also mix and match between vendors.

Erik: Good. Great. Okay. That's a trend that we've seen in a number of other industries. I'm thinking about industrial controls, where I feel like the incumbents are at least over maybe 20-year period. We're kind of fighting tooth and nail to defend space. How do you see that rolling out in this particular market? Do you feel like, in general, the incumbents are accepting this trend and embracing it to some extent, or is there also a bit of a, "Okay. We have to go with the trend, but we're still going to try to defend territory when possible"? Because, of course, companies would, to some extent still, I imagine, like to have capture customers when feasible.

Ross: What I see happening is, as that connectivity becomes cheap and ubiquitous, where you add value has to be further up in that value chain. Where in the past, maybe your value was, "I can build a wireless network for you." That might be a huge value-add to the end customer. That's kind of table stakes now. So what I see happening is, I would say, the more forward-leaning customers in that space say, "I don't need to put my engineering resources on how do I connect the things together. I'm going to put my resources on how do I mind actionable insights from all this data that I'm collecting, versus I'm just going to collect the data for you. I think that the more forward-looking companies have said, I can take my valuable engineering resources, and put them on these higher value activities, versus building yet another wireless standard to connect my stuff together. Let the market take place or take the role of connectivity, and I'll bring the value to the table on using the data.

In the metering space, there's a new wireless standard we support. Well, it's not new. I call it the largest IoT network you've never heard of. That's standard called Wi-SUN. That's an IEEE open-standard. Its genesis was in the metering space. That was the whole idea. It's that you can be vendor interoperable now. We see that replacing those proprietary networks. In the IoT space, you can look at things like WirelessHART or some of the other standards that have come along where, again, the connectivity is standards-based in multi-vendor and ubiquitous. Now for those customers, it becomes, how do I show value on the data that I'm mining from that I've collected? So that's really the trend I see. It's trying to move up the value chain and not waste your resources or spend your valuable resources re-engineering what the market has already solved.

Erik: Okay. Great. Well, that makes sense. It sounds like that's, to some extent, where you're moving with the Series 3 platform as well. I think this topic of edge or intelligence at the edge is a topic that's been on the table for 5, maybe, even in some cases, 10 years now. But really, right now, it's starting to become adopted at scale in a lot of different environments. You have a lot of companies working on being able to put relatively sophisticated compute capabilities in small footprints. So there's a lot more application space to play with there. Tell us a bit more about what that looks like. Are there a small set of initial use cases? You'd mentioned that it's important to have. And if you're early movers, that basically finance to some extent the R&D to actually get this out to market. Then there's going to be a long tail of other use cases. But what would be the early adopter use cases where you see room for scale here?

Ross: A lot of what we're seeing is that intelligence at the edge, that means that if you take a step back, you have — one of the interesting challenges you have on the edge is, you have much fewer resources than if you push the data to the cloud. In the past, you may have been looking at, I have 32 kilobytes of memory. I have a core that's running at 50 megahertz. I have some limited IO I can do. So that would really constrain how much processing you can do at the edge. You also have a limit on bat power if you're battery powered. you can't really be hungry in terms of the power consumption that device has. So you have these really constrained resources there. What we've seen over time is the ability to squeeze more out of that same kind of power size budget, being able to do more. There can be some real important drivers of why you'd want to do more on the edge. Sending data to the cloud can be expensive in terms of bandwidth, power consumption, all those types of things. So if on the edge, you can compress that data down, if you can make some insights on that raw data that's coming in, or even actions at the edge, that saves you from having to send a lot of data to the cloud. That can be a power savings for you.

There also can be privacy concerns. If you're using a camera in a room, for example, to try to do occupancy sensing, determine that people are in the room, you may not want to send raw data streams to the cloud because of privacy concerns. So if you can actually make that intelligent decision at the edge without having to send that data up into the cloud, that's so valuable for a lot of applications. I think there are a lot of factors that are pushing more processing, more intelligence to the edge. When you do that though, there are several challenges you have to overcome. I'll start at the very bottom. You want these edge devices to be secure. Security becomes a big part of the story on the edge. How can you make sure that someone doesn't get on your network and corrupt your network or take over your network in a malicious way? So if you think of these industrial networks or a smart city network, if someone could get on to that network, you could do some real damage. So security is incredibly important.

As we've moved forward from our Series 1 to our Series 2, to our Series 3 platform, we continue to up the bar on security. We have things like Secure Boot, so you know you're booting from a good firmware and image. We have the ability to verify that the software that you're loading in from the flash is signed and valid. Encryption, tamper detection and prevention — all that is important to make that end device secure. And as we move from those series, that security has continued to ramp up. Sorry. Series 3 has now the highest-level security that we've ever built. Also now, if you talk about doing more compute at the edge, you need more raw horsepower for doing that. We're starting to look at capabilities like having dual cores in these devices — one core that would run the wireless stack and take care of the communication, but one core you would give the customer to allow them to run their applications. So more powerful compute, more compute with multi core and then more memory and IO, so you can do more powerful applications there as well.

Then having that whole suite of wireless connectivity, I mentioned earlier, is really important. Certain devices will have really low data rates. Maybe you're looking at Bluetooth. Others may have video stream they're trying to capture, and you want Wi-Fi. So securing the edge, having the compute capabilities to do that first level of data processing, and then having the right wireless technology for whatever your application is, that's what we see is foundational for the IoT to take the next step and what we're trying to unlock with our Series 3.

Erik: Yeah, the security topic is an interesting one. IoT devices have been notoriously susceptible in the past, right? Often security has seemed to be a bit of an afterthought. It looks like there's been some progress in terms of regulating security in, allocating responsibility. That's a good step. But of course, security is one of those interesting spaces where it's not just a continuous line of improvement like it is with a lot of technologies. But it's a bit more like a war, right?

Ross: Right.

Erik: You launch new innovations, but the black hats are also innovating and investing resources. Where do you see this trending for IoT ONE right now if you look at, let's say, the sum? So you're making a lot of innovation. Other companies are making innovation to secure devices, but there's also a lot of investment going into new technologies to hack devices? Do you see us trending in a positive direction, or is it a bit more treading water as these two forces compete in the IoT space?

Ross: Yeah, I think you've summed up the core challenge. It's an ongoing battle. It'll never be over. The IoT end devices have been probably behind the curve in terms of really adapting to security. Some of these new technologies I mentioned earlier — we talked about Wi-SUN or we talked about Matter — there's a requirement now that every device has to have a certificate injected into the device before it can join the network. So kind of a known good certificate that you validate, hey, this is a good actor that's trying to join the network.

Previously, either the IoT standards or the devices customers used didn't really have that requirement or those capabilities. So I mentioned that Wi-SUN network. You have to inject the certificate into any Wi-SUN device that's going to join the network. The home standards even like Matter have similar requirements on the certificate injection. I think definitely the bar is getting raised as we look at how IoT is involved now to address those threats. In addition to that, there's physical tampering of devices to try to get access to them. You're seeing now in our Series 2 family, we put in a lot of that tamper detection and protection. That if you are doing a side channel attack, that we protect your security keys. We ensure that we scramble the currents that we're using, so you can't look at the current signature on a voltage pin and say, "Alright. I can see what's being pulled out of the flash." All those types of core security mechanisms now are built into our IoT devices. So we definitely have raised the bar, I think, significantly where we were a few years ago. But as we move forward, like you said, that's going to be a never-ending battle. But we have a technology we call our Secure Vault technology. It's really putting together all those pieces so that customers can build a much more secure IoT device than they've been able to in the past.

Erik: Okay. Great. Thanks. So it sounds like these are two important developments: FG28 and Series 3. Anything else that you wanted to touch on today in terms of either new projects that are coming to market or otherwise new innovations, new developments, coming out of Silicon Labs?

Ross: Yeah, I would say the one thing we didn't talk about that I'd just like to highlight a little bit is, at the end of the day, all of this technology that vendors like ourselves are bringing to the market doesn't really matter unless developers can take advantage of it. As these devices get more complicated, as the use cases get more complicated, as the networks get more sophisticated, you have to also make it easy for developers to take advantage of all of those improvements that you're bringing to the market. So we focus a lot also on that developer experience so that our customers can have an easier journey. We focus a lot on making it easy to migrate between those families I've talked about. That a developer who's developing on our Series 2 family, if they want to move to Series 3, the tools and the libraries that they've gotten familiar with on using on Series 2, they can migrate easily to Series 3.

We're also trying to support whatever developer tools developers want to use. If you look at that, we've had in the past our own IDE and development environment that has been really useful for us. But there are some standard tools out there like Visual Studio Code that is a widely-used tool in the development community. We now have opened up our products and our software to work with that, which is with Visual Studio Code. Because that's developers want to develop in, for example.

One of the things we've really focused on is that developer experience. So it's not really valuable if developers can't take advantage of all this technology we've talked about. So we make it scalable between those different platforms I talked about — Series 1, 2, and 3. Developers can use the same libraries and tools to migrate between those platforms. So we try to make it easy for them to choose the platform that's their best fit. In addition, developers have certain tools that maybe is right for them or they're familiar with, kind of industry-standard tools. Now we're opening up our development environments to also plug into those tools. So if you think Visual Studio Code, that's a very common tool that developers will use. Now all of our products will plug into Visual Studio. So we're trying to make it easy for customers to develop on these products.

Erik: Great. That's very important. I think one of the big trends in terms of development now is allowing people that are in business, people that are in operations who might have technical teams and colleagues who are not necessarily sophisticated or a large team or a startup for that matter to do more sophisticated things so that they can start to tinker around with solutions, and prototype, and so forth. And so these types of developer tools are very important for also enabling that activity so that people that are frankly closest to the application but maybe farther away from the core technology can be more involved in this process.

Ross: Right.

Erik: Ross, thank you for the update. I think a great set of innovations that are coming to market. The timing. Maybe last question. The timing for these, when will they be available?

Ross: So we're sampling our FG28. We'll be launching that in the back half of the year. Then you'll start to see products from us on our Series 3 families a couple quarters beyond that.

Erik: All right. Wonderful. Well, Ross, thanks for the time today. I really appreciate it.

Ross: Thank you, Erik. Appreciate the opportunity.