Erik: Welcome to the Industrial IoT Spotlight, your number one spot for insight from industrial IoT thought leaders who are transforming businesses today with your host, Erik Walenza.
Welcome back to the Industrial IoT Spotlight podcast. I'm your host, Erik Walenza, CEO of IoT ONE. And our guest today is Kipp Jones, Chief Technology Evangelist of SkyHook. SkyHook provides customers with real time services and analytical insights via a combination of precise device location and actionable venues. In this talk, we discuss the key variables that determine location requirements for a device, and that guide technology architecture decisions. We also explored the capabilities and limitations of new communications technologies such as 5G and LoRa.
If you find these conversations valuable, please leave us a comment and a five-star review. And if you'd like to share your company's story or recommend a speaker, please email us at team@IoTone.com. Thank you. Kipp, thank you so much for joining us today.
Kipp: Well, thanks for having me on. I'm excited about this.
Erik: So Kipp, before we get into SkyHook, and your business model and your technology domains, I want to learn a little bit more about yourself so that the listeners know who they're talking to. You've been in six or seven startups before joining SkyHook, I think you have 19 patents to your name. Can you just quickly walk us through the path that you took now to be the Chief Technology Evangelist of SkyHook?
Kipp: Yeah, I did my Master's in Computer Science at Georgia Tech and ended up connecting with who would then become the founder of SkyHook, actually started with working with him in 95. So I had gone through a number of startups and continue to enjoy that more entrepreneurial tack on things. So, when I went back to work on my PhD, all but dissertation on that, but I ended up working with SkyHook during that time period, that's 2003-2004 timeframe when they were getting things rolling. So I worked remote for a while for very early years with SkyHook. So it's been a long, almost like a startup still continues in that sort of vein just because the industry changes so fast. It continues to feel like every years a new start.
Erik: And I see that right before you joined SkyHook you got through two years of a PhD over at Georgia Tech and you just decided what SkyHook was doing was too interesting and you wanted to get back into the operational side of it?
Kipp: Yeah and just life in general. At some point, you have to make some money.
Erik: SkyHook, maybe starting from the value proposition, what is SkyHook bring into your customers?
Kipp: Sure. The basic premise we started with is location. Early years, as I mentioned 2004, really when we started things up, it was, hey, all these devices are going to need location. GPS is good. But there are cases where it is not that good. And we know that there are other challenges with GPS. And so we started figuring out how to make WiFi positioning work as a positioning technology, and then figure out how to make that scale to a global system. So our proposition really is we can flexibly determine location in nearly any device that has basically any radio and determine that the position of advice. So the basic premise is people like to know where their things are, whether it's for personal use, or business optimization, or finding your way to the ubiquitous SkyHook.
Erik: And are you normally looking at larger more integrated devices or do you also look at a system perspective? For example, we're working on a smart factory project right now where we're looking at using RFID to locate where pallets and containers and so forth are in a factory? So there it's quite different from tracking something by wireless. Do you also look at that more system perspective or is it generally an integrated device?
Kipp: It's generally an integrated device. We do a lot of work from the chip side all the way up to smartphones at app level. So there's a whole bunch of integration options. And that's one of the things I think we focus on is flexible integration. Obviously, our background is really looking at not as much RFID side of things, but more along though the WiFi, Bluetooth, cell, GPS combination, coordinated with sensors when available.
Erik: So there's a couple things that I found interesting I was reading through your website. So one is you have this tagline “SkyHook provides customers with real time services and analytical insights via combination of precise device location and actionable venues.’ What is an actionable venue?
Kipp: So, really that the aspect of now shifting towards the data analytics side, and the use of said data for optimization and other use cases. So, actionable venue is really the idea that you can apply this data to specific points of interest. So if you're a retail company, now we can do the analysis to determine your foot traffic and whether that's going up or down. In today's view, you can use that examine how effective and what the effects are of, for example, regulations on COVID stay at home orders and such.
Erik: Let's say, we're talking about a jewelry store, and they have a due demonstration in one of the corners, so you can tell how much foot traffic is coming into the venue? Can you also tell how many people are actually spending time at that new demonstration to get a feel for what is the marketing impact of putting that in and maybe how long do people stand there? Do you have that kind of granularity in a venue?
Kipp: We can get there. It's both business and technology challenge. Indoor, we've got a massive effort right underway right now for to continue to grow and improve our indoor positioning capabilities and scale. Indoor positioning at the level you're speaking about is generally a one off. It's going to require additional hardware in general. And you're going to need some specialized software that ensures that you are able to determine that location and that dwell time at that high precision.
So our business model has been basically we can find location everywhere. We do improve location to get to these use cases, and we can, and we have technology that will get you there. The base layer is off of what we'd call signals of opportunity. Beyond that, to increase your accuracy, you often need to go to additional capabilities, whether that's some of the new technologies in WiFi with round trip time, or Bluetooth or some other hardware to improve that positioning capability.
Erik: Okay, so then that would start getting into more of a system integration role where somebody would go in and build a customized or semi customized solution?
Kipp: Yeah. You can deploy extra WiFi. You can deploy Bluetooth beacons, for example, and improve that. But again, if you've got multiple or 100 stores or whatever, the cost effectiveness of that starts to be a challenge.
Erik: So let's go then back a little bit towards your business. So you have the retailers, the venue operators, and then you have all the device OEMs, are you primarily selling technology through to device manufacturers? Or do you ever get into this kind of integration work for the end deployment sites?
Kipp: Well, I would say yes. So our sort of distribution and customer engagement, it's a really broad variety. So everything from integrating directly with chip manufacturers so that the location is pre-certified on certain chipsets, for example, Qualcomm is one of our partners. We've got pre certified chipsets from smartphone chipsets down to IoT chipset. So those are enable that anybody who integrates with that chipset to flip on location service, so all the way up to telco providers.
So we'll work with was carrying that are deploying a new 5G network and want positioning to be built into that all the way down to the two guys in the garage or two girls in a garage building a new pet tracker. And we'll walk them through what the ins and outs are of different technologies of different capabilities, battery life, communication costs, all of the different components because we've worked with a lot of different and variety pack of devices. So, yeah, it goes up and down the ecosystem.
Erik: But in all cases, you would be selling primarily technology, technology, maybe it was some advisory around that, is your business model more or less a pure technology/licensing model?
Kipp: Yeah. The basic system relies on some components on the device to measure the radio signals and capture that data and send it over. And then our API servers that will resolve it, obviously, we can resolve on device as well. So, it's generally a combo of those two components working together. But we do custom solutions as necessary. We've gone built RTLs operating system version for people. So, like I said, it's a very flexible integration option. And that's one of the things I think we focus on is just being able to work within, not just the technology, but the different business models that you see out there.
Erik: And your website, so you have a couple of figures here. You have 5 billion WiFi access points, 200 million cell towers, 20 million actionable venues, let's say, I'm using a device that has your solution, does that mean that there are 5 billion WiFi access points around the world and 200 million cell towers that are all being used to locate this device, not all of them, but depending on proximity that would be used, is that how to interpret that?
Kipp: Yeah. Basically, the way it works is where with GPS, you have a small number of satellites. With cell, you have a larger number of cell and it's really cell IDs, so each cell tower has multiple cell IDs on it. So it depends on how many different air interfaces and bands and all of that. So yeah, if you think about that, no matter where you go, we can give you a cell location. Multicell is obviously better, but then WiFi improves that. And yes, some 5 billion WiFi access points geolocate around the globe, 5 billion that allow positioning at different scale in sell today.
Erik: So, the focus of our podcast is IoT, so I want to go into some of the use cases. But before we kind of go down that rabbit hole, you divide your offerings into precision location, and geospatial insights. And precision location, I think is quite intuitive and there's a lot of IoT use cases that have different requirements around that. The geospatial insights, I thought that was very interesting because I think people don't really think about this almost ever. You do think about maybe Uber having some geospatial insights from their fleet of vehicles and the ride sharing.
So you think about people that have these dispersed assets having some insights that they collect, but you don't really think about a company like SkyHook having the ability to extract geospatial insights from this network. And then I think there were some things mentioned around the competitive insights, certainly marketing insights. But can you explain a little bit about what would be included in this solution bucket of geospatial insights?
Kipp: So there's a number of categories. Really as you mentioned, there's a bunch of marketing insights that that companies can get. And again, we've started from a number of different avenues. So as a smartphone, traffic and things like that, there's a lot of interesting capabilities that can be provided from that whether you're a retail or sports team, although sports teams are not doing so well these days.
So really think about that marketing and sort of business, location, site selection, all of that kind of stuff. And then as you go into the IoT side, you start to look more at the logistics side of things, and what you can learn from the location data about the behavior, whether it's trying to look at equipment fail failures, what's happening; good example is hold storage travel.
So generally, you've got sensor tracking the temperature, and understanding that if your freight gets above a certain temperature, it can go bad. And so you want to track that, you want to know where that happens, if that is a consistent issue someplace, those types of things. Location, obviously, is a big component in that.
I think, one of my favorite use cases when you talk about the sort of location insights, we had a company a large manufacturer of appliances who had a new smart connected watch machine that they’re getting failures, and couldn't figure it out, couldn't recreate it. So we said, well, hey, do you have access to WiFi scan because they're connected? And they said, yeah, that comes with our diagnostics. So basically, we figured out approximately where those devices were, then we said, hey, what's the elevation of these? And they were all above a certain altitude. So they are having a part fail on altitude. And that was, again, based on that geospatial intelligence and insights, then you got the ability to recognize and not have to recall all of your devices but figure out what that part failure was.
Erik: Yeah, let's go into a few of these IoT more related situation. So, you already mentioned, logistics, or asset tracking, we now have a lot of connected cars, we don't really yet have autonomous cars, but I suppose that this would be a very necessary solution. What does that look like bcause there you have, on the one hand, a very private device, but you also have it in a very public space interacting with other devices and then with city infrastructure. What would be the location environment around the fleet of connected cars in a city?
Kipp: There's a couple of cases. The most prominent one that we've been involved with would be more on the insurance side of things: so providing additional location facilities within those solutions. There's obviously as you mentioned the autonomous vehicle, which also just integrates a whole slew of other technologies, obviously, for that purpose.
But a lot of times around destination, determination and things like that, when you go inside a parking space and lose your location, things like that, tunnels and things, there's a number of use cases where it actually provides additional capability. Most GPS is most prominent in vehicular navigation and that, obviously, they're mostly outside driving. So those don't have as many needs as other IoT solutions, but certainly one that we've had some impact on.
Erik: So if I look at connected cars, maybe if I look at wearables which are attached to a person, there's this security aspect or personally identifiable information aspect, which I know is a hot topic with governments right now. I happen to be making this call from China, so it's not such a topic here, it's been decided. But maybe in Europe, in the US, there's still more of an ongoing discussion. Do you have governments calling often to figure out how to access location data? What is the current policy around, maybe not just a policy for SkyHook, but more generally in the industry around how to manage this data when there's different government agencies that might want to access it for different purposes?
Kipp: Yeah, in general, we just work with our customers. So I think we are under the radar a lot on that. There's companies with so much more data. If you look at the Apples’ and Googles’ who just have a vast deployment, when you're looking at personal data or your smartphones in particular read, which is where a lot of that data comes from. It's a little less, I guess prominent, and likely to have enough data to provide that in many cases.
As far as these mobile personal emergency response systems, impulse category, the health tracking devices, we do a lot of work with those types of devices. Also, interestingly, we do a lot of work with the home detainment devices. So if you need to detain at home, we can help provide location for that purpose. So on the data side, it's not as prominent, like I said that the core of our business is computing the location.
Erik: So if they want to access data around a particular device, they won't contact you, they would contact the device manufacturer or maybe some higher level component?
Kipp: Yeah. Within the operating system of Android, and iOS, there's a lot more location happening in those.
Erik: So let's go a bit deeper now into the technology behind SkyHook. You have, according to your site, about 650 patents, so quite an extensive patent field here. What are the different areas of innovation that you're focused on?
Kipp: Well, so a lot of that there's several families and classes. So if you look at the technology directly from SkyHook, our growth periods, it's all around WiFi massive scale, WiFi positioning, everything from computing the error estimate to deriving the data to positioning algorithms themselves. So, a lot around Wi Fi positioning at scale, that's a huge chunk. And again, there's multiple families within that.
Also, we integrated with another company called TruePosition, which had and continues as a combined company, extended our cell capabilities. They had done a lot of work and cell positioning, different UTDOA type of technologies, a lot of patents in that arena as well. So we scale really well equally across the WiFi and cell positioning technologies. I'd say that's where the majority of them are, again, in positioning technologies.
Erik: So 5G is a big topic right now, a lot of money being put behind the deployments. But I suppose we're probably not going to see a huge number of devices be 5G compatible for at least maybe a couple years. But is that the next field where you'd be carving out territory? Or how important is 5G right now for your research efforts?
Kipp: It's way up there. We are 5g capable now. So as they roll out, we are both collecting and providing location off of that data. We also look at the newer LPWAN type of networks as well since those will provide really for that low power in IOT space, that's probably a place where a lot of devices will live. We have to make sure that we're capable to produce and create location for those types of devices. So we've done a lot of work in that space as well to accommodate the market shifts. But yeah, it takes a while for these technologies to roll out to scale. We have to start early on those to be ready for it when it happens.
Erik: So if you look at 5G, LoRa, NB-IoT, Sigfox, from a location perspective, in terms of what you're able to glean the maybe precision or latency or actually, maybe I should ask what are the major variables, but what are the differences in terms of your ability to access data from these different communications technologies?
Kipp: So yeah, the different ones offer really different capabilities. Their power deployment, with LPWAN, you're trying to do a lot more longer distance travel on the tower so the accuracy of the location can degrade. The variety of cells that you have can degrade the location capabilities. One of the things we've seen a lot and we've worked with a number of these providers in the LPWAN space to do integrate WiFi collection through their network, and sometimes that's a protocol issue to provide that low power.
Also, they care a lot about bytes at some time. You think in today's world, bytes would not be a thing that you're counting. But in those, they do. So there's also issues about how to get a WiFi scan into 12 bytes. So a lot of times what we see is you can do a partner WiFi chipset and LPWAN. You can do a scan on the WiFi and ship it over the cell based communication. And that seems to work really well. You can control the different technologies in your application or the iOS. That's one way to integrate and improve location capabilities beyond what the network can provide directly.
Erik: So you have the network, you have WiFi, you have 4G and LPWAN, and these other communications technologies, and then you have the devices themselves, which create a lot of additional endpoints that might have location data and could allow you to be maybe potentially much more granular. If you know that this device X is in proximity to device Y, is that possible to do? Is that an area that's interesting at all to look at, I don't know, it's not really a mesh network, but somehow, using the IoT devices themselves as location points to more finally identify the location of your desired object?
Kipp: It's in a collaborative positioning space. It is possible. It’s feasible. The biggest challenge is motion. With most devices, you can compute that are static, that are stable, if it's moving, you need to know how fast it's moving and so you have to have a high frequency of observation in order to make that useful. So using sort of mobile phones or mobile anything, as a positioning, GPS works really well because they know exactly where those satellites are at all times. They're mobile, but they're not randomly moving. So the difference is that. So yes, you can do that. But it's a tricky thing to work at scale.
Erik: Anything else, because I am so far for being an expert in this particular domain, so there's probably questions I should be asking. But around the technology domain, anything else that's really critical to understand that we haven't covered?
Kipp: Well, I think there's a couple things. One, I would talk about the different layers you can integrate. So as I mentioned, you can think about this, getting it with your chip. You can think about getting it with your operating system. And think about if you are building a brand new device, and you're choosing an OS that doesn't have built in location, you can build it into the application layer. So there's a lot of integration questions that we get that we work with people on. So that's one aspect.
I think one of the things that we work on as well is I mentioned integration options. When we look at feature phones, which are really today's flip phones basically, they generally have a pretty robust operating system, but oftentimes it's not with location. So we have options to integrate into those.
Also, from regulatory E911 is something that drives the industry and we're always coordinated in with a lot of that. We've done some SIM and eSIM integration which I think is pretty exciting so that they can operate at an emergency level. I think from a business perspective, you may or may not have to get down into these weeds, depends on whether you're building a device or you're using just taking somebody's device that's been built. I mean, you kind of need to know its capabilities, but you don't need to know how to integrate it.
Erik: So from this regulatory perspective, if you're a device manufacturer, and you want to sell in 150 countries around the world, is there a significant amount of complexity in terms of what location device you're able to access in different regions? Or are we fairly well unified across the key data points that would be needed for most use cases?
Kipp: Yeah, I mean, pretty much available in protocols. There's a lot of protocols available for this, but it has to be implemented. So I guess it's capable. I know a lot of the FCC regulation require new devices that come online to be able to support for example, these E911 restrictions, so they have to pass that. Those are areas where we see that pushes the industry straightforward and accuracy and capabilities.
Erik: Maybe even before we go into a case study, you can just start with what would be the first set of conversations that you would have with a potential customer? What would you need to know in order to know how to work with them?
Kipp: Well, let's take like a company that is building either a health tracker or a pet tracker, for example, so some kind of IoT device that is meant to provide some location capability and communicate periodically with some back to home. When they're building these devices, the first we often come in after their first generation. And it's getting better now as far as knowledge about location and location capabilities in the market for device manufacturers.
Largely, what we saw early on is you would get a GPS chipset, and build out your device and roll it out and then determine that, hey, you lose your battery faster and when it goes inside your dark. So often, we work with folks and say, well, what chipsets do you have? It's a lot about finding the requirements of battery size, cost, accuracy, along those various axes to determine what the best solution is.
Cell communication, for example, if you're tracking a truck across the country is generally good enough when you're on Interstate. But if you want to go into city, then you're going to probably want more accurate if you want to track the route and things like that. So depending on those use cases, and then if you need to go indoor, where your assets going to go indoor, or your pallets going to go indoor, you need to be prepared for that.
Erik: What is the kind of the extreme of precision that you're able to accomplish in a typical situation based on WiFi or cellular?
Kipp: With WiFi, it varies and based on density of access points. So you can think of sort of the dense areas, you're going to get 15 meter, 10 meter, 20 meter or somewhere in that zone, then as you get into more suburban area, you can lose some of the accuracy because they're not as dense, so you have fewer access points, fewer data points to compute. So, you can go to 50 and then you get out there to the 100, 100+ if you're really in rural. Really if you think about cell, I generally think of 100 meters as a good point. It can get down to 50 in dense urban again. So as you go out to the country, though, you're looking at kilometers. So potentially, with the newer technology, generally trends towards more accurate location.
Erik: Because with Uber feels like you're able to get within a five meters or quite close to the location is that what technologies would be underlying that accuracy?
Kipp: So a lot of GPS for Uber. They use WiFi as well to augment that. If you look at GPS in Manhattan, for example, you have large error just because of the urban Canyon echo thing and bouncing and signals. So often that there's a hybrid approach, which improves the location. So you can use the WiFi data to help determine when your GPS has gone awry and correct for that. But they also have integration with dead reckoning, so you can use your sensors to integrate that in your motion so you keep a consistent thing. They do a lot of road map matching as well, right, which is going to say you're probably on this road so we can snap you here to make it look stable and so you're not bouncing.
Erik: And then, so we'd have this set of questions that we'd be asking. Who would you be talking to? Would you be speaking with the R&D team? Would this be with more product development? Or on the business side, who would be the people that you initially engage with?
Kipp: Yeah, generally, it's the product owner who's got this challenge. If we carry on right, the first iteration of the product we had some challenges with the location. It wasn't satisfactory for the use case, so the product person is trying to figure out how to solve that. So generally, they'll initiate and a lot of times, there's an engineer certainly involved to do that. But it's brought on by the product person generally.
Erik: Yeah. I mean, it seems fairly technical for a product person, but I guess that's why as you said, many first generation devices go out without solution because people don't think about it beyond?
Kipp: Other times they're really focused on getting it out of market and getting that feature that they want which is for the cat feature, right, then they get into the actual nitty gritty.
Erik: So let's say we have a healthcare device, you've answered the background questions. What would be then the path from these initial discussions towards implementation?
Kipp: So then it's just a matter of which chipset which operating system you have. Where are you going to integrate it? Are you going to use one of our SDK? Are you going to implement it on your own and use just the API? Obviously, we have spent a lot of time on the process of gathering the data and packaging it and computing it and all that. So, we like to press our technology and just because again, we've solved this problem lots of times.
Then it's working with the engineers figure out how best to integrate. So based on all that we'll work with them to figure out is what's the best solution that we can offer them. And then we'll work through it. If we have to work with them on a custom integration, we will. In fact, we've done the work with Philips on the health care, basically, the wearable dongle that tracks where you go, and if you fall down. So we've done a lot of work in that area. In fact, that's a great one to think about for battery life, where you don't really want to, and you spent a lot of time inside. So you don't necessarily need to have GPS turned on, except in the case where you fall down, and then you want as much information as you can get battery began.
Erik: So let's say the sensor indicates that somebody has fallen down, then you can kind of activate these other data sources that will lead to the battery, but give more precision?
Erik: And then the pricing model behind this, I guess, you have your API, so you have a service built behind the access to the data, you have maybe some integration work that needs to get up front. Is it like per device based on maybe the amount of data needed for per month? Or how would this type of solution be priced?
Kipp: It is dependent. A lot of times, so if you're a major manufacturer of these, they like to bake it into their bill of materials. So it'll be a unit cost basically built that's going to last for two years, or whatever the lifecycle is. So that way they can build it in their bill of materials and ongoing variable costs with it. And if you are doing a different service, or you're not at that scale, you may choose to do a transaction-based model.
I'm not a fan of transaction-based models, just because I feel like that creates a barrier to using location. Again, like our business model integration is pretty flexible as well try to work within the scope of what their business model needs. Obviously, we need to generally get paid as well, but we’ll work that in.
Erik: So Qualcomm, I guess, in a lot of cases, it would just be built into chipset, and then if somebody already has their hardware, and they want to build this on as version two without changing in underlying hardware, they might have to use a different solution where it's not baked into the hardware. Can you give us a rough estimate for what the cost might be? I mean, are we talking about cents per device? But what would the Because you mentioned there's a preference to kind of bake it in as opposed to letting the device OEM try to control access to location because they want to save pennies. But what would that actually end up looking like per device over a year or so?
Kipp: Yeah. Device, it is just an appliance that is just every once in a while checking in. That's completely different than a mobile smartphone scenario. So it will vary based on expected usage. Honestly, that's one of the drivers of our costs, is how much our API servers are getting beat up. And again, it varies. But it's way less than a chip that you're going to pay for. So you can figure somewhere under that.
Erik: So you mentioned a lifecycle, so let's say Qualcomm bakes into the cost of their bomb for assuming kind of two years, but then, maybe the typical life cycle in the US is two years and then after that the device is sold to somebody in India who uses it for another five years, will the functionality cease after a particular time period? Or how would the life cycle look like if it extends beyond maybe the planned duration?
Kipp: We're pretty gentle on end of lifing. It, basically, from accounting standpoint have to have some end of life commitment. In general, we will work with folks to get the right lifecycle expectations. Obviously, we won't just shut off the customers if a device last longer. So generally, you're looking at like a three year life cycle on a lot of this stuff. But yeah, if it goes on, it depends on if it gets rebuilt as something or if it's just continues to be used. We'll see those go on for a long time. So you'll see some death on those devices and so you've got a tail and potentially that'll go on. We can turn them off, let them survive.
Erik: And I guess it's a probably somewhat of a normal distribution, where there's some devices that somebody loses after six months, and then there's other devices that go on for six years, and then there's an average that you can press on. So I think this is super clear. Is there anything important that we missed, maybe around the implementation? Any other critical considerations?
Kipp: No, like I said, I think, just from recommend people do a little research on location and not just take the manufacturer or what comes with it, just make sure it meets your needs.
Erik: I think location is getting more important than ever. We're more focused on the industrial IoT space. And location is so critical to enabling business models and use cases. So certainly, I for one, appreciate that.
Kipp: Exactly why we're investing in that space.
Erik: Great. Well, Kipp, thank you so much for taking the time to speak with us today.
Kipp: I appreciate it. And I look forward to working with you in the future.
Erik: Thanks for tuning in to another edition of the industrial IoT spotlight. Don't forget to follow us on Twitter at IotoneHQ, and to check out our database of case studies on IoTONE.com. If you have unique insight or a project deployment story to share, we'd love to feature you on a future edition. Write us at erik.walenza@IoTone.com.