ITA's IoT Summit Asks, "What's Holding Back the Internet of Things?"
Thursday, December 13, 2018
Posted by: Kaylin Berg
Forte Q&A: Founders of edge-computing startup Edgeworx discuss the role of AI, blockchain, the edge, and the cloud in shaping the future of the Internet of Things.
The Illinois Technology Association (ITA) kicks off its sixth annual IoT Summit today, where Chicago thought leaders and innovators will discuss the future of the Internet of Things (IoT).
The theme of this year’s summit, “Pilots to Production,” will explore why so many organizations see the successful launch of IoT products but struggle to scale those successes to full production.
In advance of the event, we spoke with two IoT thought leaders: Kilton Hopkins and Farah Papaioannou, co-founders of Edgeworx, an edge computing startup that recently announced funding from Samsung NEXT. Edge computing is the practice of processing data near the “edge” of a network, on or near the device itself. Edge computing enables data produced by IoT devices to be processed at the source of that information, rather than sending it to a data center or cloud for processing.
What follows is our discussion with Kilton and Farah on edge computing and other IoT topics that will be addressed at this week’s summit.
What do you see as the biggest inhibitor and enablers in progressing IoT within enterprises today?
Kilton Hopkins: One of the biggest enablers for progressing IoT is the push for companies to obtain more and better data. Organizations need that data to do things like enhanced services, predictive maintenance, and real-time analytics. If companies want to remain competitive, IoT will become mandatory to obtain that data. Four years ago, companies might have been able to stay ahead of the competition without this data. Now, if you don’t have the data, you’re falling behind.
Across all these different industries, the same inhibitors keep coming up. The inhibitors are universal, but in different industries, they’re different things. In the oil and gas industry, for instance, it’s called a “truck roll,” meaning they’re sending someone out in the truck to fix what’s happening with the digital technology at the oil pump. But then in, say, commercial real estate, they may not call it a “truck roll,” but it’s the same idea: Someone on foot is going around fixing things.
The basic concept is that you’ve got these things that prevent scale. It’s not that it prevents the proof of concept (PoC)—it prevents them from growing it.
So, companies are getting stuck in the pilot phase instead of moving to the production phase. What else, in addition to a lack of automation, is holding back production?
Kilton: One of the reasons that companies are having difficulty moving past the pilot phase is that the systems they used to set up their pilot were never designed to be operated at scale.
Think about it this way: If you use Amazon’s AWS [Amazon Web Services] IoT to do your pilot, it allows you to buy some ready-made sensor boards, then you can install the AWS SDK [software developer’s kit] into your boards. That gives you some immediate functionality and your data starts flowing.
You might say, “Okay, this pilot has been pretty cool! We have data now from our IoT sensors, we’re analyzing that data, and we did it for 20 machines in the factory.” The boss loves it and asks what it will cost to roll out to 10,000 machines within the company’s factory system. The cost of the hardware? $1 million. The cost of labor? $10 million. That’s because the AWS IoT software and infrastructure platform takes so long to set up a single device that you’ll have employees do nothing but set up sensors for an entire year. That doesn’t make any sense. It’s why people are having trouble going from pilot to production.
To move forward, we need systems where pilots can be easily created, but then production can be scaled with the push of a button. That’s one of the things we focus on at Edgeworx. We realized years ago that if you can’t find a way to roll something out easily and cheaply, it’ll never get rolled out and the project will get canceled.
You mentioned a couple of industries that understand what the issues and opportunities in IoT are. Which verticals “get it”? Which verticals struggle?
Farah: Edge computing is a new term and people are just starting to invest in it, but edge computing has been around forever. Your mobile device is an edge device: It computes at the edge and takes data from your surrounding resources.
The verticals that have been doing it for a while—oil and gas, for instance—those guys are far ahead of the curve. In others, where you’re starting to see adoption, like what you might find in a smart home or autonomous vehicle, they’re a little slower, but they’re equally as excited and they’re going to get there in their own time.
Kilton: Another one is the telecommunications industry. The reason they’re adopting edge computing and IoT faster than others is that they’re at the center of the data flow for everybody, everywhere. So, whether it’s the fiber backbone or the cell towers, providers like AT&T or Verizon are that center point for the data flowing through. It’s a golden opportunity for providers to implement IoT subsystems, and especially edge computing. We see that as one of the key reasons that they’re picking up faster than the others. They’re either going to be a commodity pipe, where peoples’ data flows through and they don’t add any value, or they can modernize themselves with new revenue opportunities before they become obsolete.
The oil and gas industries are adopting faster because they must. If they’re going to survive in an era of increasing costs (the cost of human beings keeps going up and the price of oil keeps going down) the only way to survive is if they can find a way to stop sending a person out who costs $400 an hour to fix every problem. They can’t suffer it anymore, and they’re adopting IoT and edge computing to solve it.
AI and machine learning are at the center of the IoT Summit agenda. Are these just the buzzwords everyone is using, or do you see each as a fit for IoT advancement?
Farah: AI has been a big driver for what we’re trying to do at the edge. Everyone wants to collect more data. One of the challenges is that when you collect more data and do more analytics, you must send all that data back over a network to some sort of cloud. There’s a huge cost for the bandwidth, latency, network connection, and security associated with that transfer of data.
For example, in autonomous vehicles: if you find a few cars at a four-way stop, there’s some intelligence required to decide who goes first. With the way things are set up today, you send it over a network to be processed in the cloud. That transfer—sending the data, analyzing it, and sending data back—creates something like a 250-millisecond latency. Well, the cars aren’t going to wait that long. There’s going to be some sort of accident.
With edge computing, you can run those algorithms right there at the edge, on the device itself. There’s no need to send some data back to the cloud. AI is driving what we’re doing at the edge.
When that data is analyzed on the device rather than in the cloud, does it increase the hardware cost of each device? How does that cost influence edge adoption?
Kilton: Depending on the level of AI, your increase in hardware costs could be as low as $10 or as high as $1000. Somewhere around the $500 mark, you get almost as much capability as a cloud server has. That’s for the most extreme cases, like real-time video analytics for a security video, for instance.
It varies. The hardware cost increases can be substantial, but in most cases, there’s a real ROI. So, you’ll invest in the new hardware and the savings will pay it back within a few months.
Farah: With most of the customers Edgeworx is working with, we’re running on existing hardware that’s out there. Often, you can add additional capabilities via software rather than hardware.
Another hot topic: Blockchain. What do you see as blockchain’s role in IoT?
Kilton: It’s mostly been used for experiments. But then, if you think about recording really important information, yeah, you can record it in the blockchain, but public blockchains aren’t going to be the key enabling technology that makes it happen. Anyone who says IoT and blockchain together is the perfect combination should probably try to do IoT first, then see how much they really need blockchain.
One of the primary values of blockchain is that you can store information that’s accessible to all different parties, it can be verified, and it’s immutable. Once it’s recorded, it can’t be changed. That’s valuable for audit trails, supply-chain transactions, verifying the authenticity of medical equipment—all these are great uses for blockchain.
But the truth is, there are other ways to store information with that sort of reliability. Even if you do use blockchain, very rarely do you need to record that stuff in the blockchain with any kind of a real-time framework. Most of the time, it would be enough to summarize a whole day’s worth of transactions in the IoT system and record a fingerprint in the blockchain for that may be daily or hourly.
So, the hype around blockchain is that folks talk about how blockchain is going to give IoT some kind of real-time, immutable record of everything that’s happening. However, the current blockchain can’t support that—and the private blockchains don’t have the security. There’s not really any situation we’ve ever come across where someone absolutely needs blockchain as the technology for IoT.
I like blockchain, I just don’t think it’s that valuable for this situation.
Do you think panelists at the IoT Summit would agree?
Kilton: My guess is that panelists that are part of a blockchain-related company would disagree by pointing out the amazing capabilities of their blockchain system. But I think larger companies would agree: We’ve yet to see a scenario where blockchain is necessary. But I’d bet those that are promoting blockchain would give 1000 reasons why you have to have it.
What opportunities do you see for companies to partner with platforms like Edgeworx?
Farah: We love organizations like Forte Group and other integrators. Our business model is largely predicated on those partnerships. We’re a completely horizontal platform. We like to think of ourselves as something like, “Android for the edge,” in that we provide you a platform for developers to make it very simple to deploy code on any piece of hardware.
We like the integrators for several reasons: One, they have the subject-matter expertise. Kilton and I don’t have the specific subject-matter expertise, nor do we want to. We do not want to be in the business of writing applications. We want to be the platform that serves those applications; we want to be the infrastructure for the edge.
A lot of SIs [system integrators] come to us and say, “Hey, our customers are saying we need to do things at the edge. We don’t know how to do that.” In some cases, they’ll tell us, “We tried it ourselves. It’s very, very difficult.” Edgeworx offers that technical capability, which allows SIs to focus on integration and subject-matter expertise—neither of which we want to focus on. So, it’s a great partnership.
We’ve worked very closely with SIs [system integrators] in our partner program. We’re enabling them to be successful at the edge, and we’re making it very simple for them to do that. It’s a huge part of our strategy and allows us to stay where we want to stay, which is providing the infrastructure layer, providing that software layer, and enabling everyone to be successful.
This allows organizations like Forte to focus on designing and developing custom apps without starting from scratch with a new toolset. Developers can build using their existing toolkits, which makes it very simple to deploy using any code or framework, using any language, and using any SDK. Integrators can build on the Edgeworx platform and businesses can focus on what they do best: Running their business.
Kilton: This is essentially how we’re going to be able to have our technology help across every vertical: The people that already have the industry relationships and the know-how.
Where do you think IoT will be in three-to-five years?
Kilton: I love making predictions like this. Now let’s see if any of them turn out to be correct! In the next three to five years, people will be saying that open-source for your infrastructure is the only thing that makes sense. That means that we’ll have more and more organizations moving away from their private blend of operating systems and specialty code. They’ll start to look to open-source as being the foundation of their smart devices.
If you think about it, if a company is making, say, home security cameras using IoT technologies, they’re required to have their own cloud and their own software on their security devices. Unless they’re the best team out there, eventually someone using open-source is going to beat them, because open-source is being cared for by the whole world. There’s no way a single company will beat all other companies. So, open source wins.
Another prediction: Right now, we have real problems with data. Ownership of data, privacy, and protection and security around data. We’re talking a big game. Everyone’s talking about security being the number one issue on their mind, especially regarding IoT.
People are well aware that there’s no security in IoT, but they’re not doing anything about it. They’re not doing anything about protecting data, they’re just saying that “we need to do this.” Data provenance—the concept of knowing where every piece of data is, and where it’s been sent from generation all the way to retention—is a serious concern right now. In three to five years, it won’t be a concern. It will be a requirement.
Companies will say, “we can’t use your product if you can’t guarantee us the location of our data at all times.” That’s something we specialize in at Edgeworx, and it’s difficult to do. Most companies are kicking the can down the road because they don’t know how to address the data problem, but they’re getting a little scared because there’s no security and there’s no data control.
Security, over the next three to five years, will be the major issue.
This post originally appeared on the Forte Group blog, here.