In episode 39 we speak with Brett Murphy, Senior Director of Market Development. Brett discusses why IIoT is one of the hottest topics in industries today. Additionally, Brett breaks down his favorite industrial IoT use cases that require real-time data exchange in distributed environments.
In Episode 39 of The Connext Podcast:
- [0:37] How DDS plays a role in IIoT
- [3:04] Brett’s favorite IIoT use case and why it’s so interesting
- [7:40] Where is the low hanging fruit for IIoT and what use cases will deliver the fastest ROI?
- [12:20] Use cases that will have the biggest impact and transform the industry
- [Blog] What is IIoT? The Industrial Internet of Things Primer
- [Blog] Connext DDS and the Industrial IoT: The Top 5 Things to Know
- [eBook] The Rise of the Robot Overlords: Clarifying the Industrial IoT
- [Podcast] Clarifying the Industrial IoT
- [Product Page] Connext 6
- [FREE On-Demand Webinar] The Rise of the Robot Overlords: Clarifying the Industrial IoT
- [FREE On-Demand Webinar] How the IIC's Connectivity Framework Guides IIoT Connectivity Selection
- [Whitepaper] A Converged Approach to Standards for Industrial Automation
Steven Onzo: Hello everybody, and welcome to another episode of The Connext Podcast. I'm here today with RTI's Brett Murphy, senior director of market development. Brett focuses on the industrial internet of things and is also responsible for RTI's initiatives with industry consortia and strategic partners. For over 20 years, Brett has worked on controls analysis, hardware-in-the-loop test systems, robotics, real-time software development, and systems engineering in the aerospace and embedded software industries. I’d like to welcome to The Connext Podcast...Brett Murphy.
Brett Murphy: Thanks for having me.
Steven Onzo: Of course, you're somebody we've wanted to interview for a while now, so we're glad you could make it on and allow us to pick your brain.
Brett Murphy: Great.
Steven Onzo: I want to start things off by recognizing IIoT as one of the hottest topics in industries today. The what and the why of IIoT has been covered exhaustively through the media and at conferences, so today we're going to focus on where the rubber meets the road. So let's talk about industrial IoT use cases, specifically use cases that require real-time data exchange in distributed environments. So firstly, can you talk about how DDS plays a role in IIoT, and how this has evolved in the past few years?
Brett Murphy: Sure. So as anybody who's listened to any of our podcasts or looked at our material, obviously DDS is a connectivity framework that enables the sharing of data across compute nodes in an industrial IoT system, from edge, to fog, to cloud. It's one of the key standards, but it's also unique amongst all these different connectivity standards that are out there in that it helps implement what we call a databus. And a databus, you can think of a little bit like a database, you read and write data to the database and the data decouples the applications.
Anyhow, technical details aside, so DDS uniquely creates this databus. And if you look at the needs of IIoT systems, they're really about the data, getting the data from the things to the analytics, to the control applications, and all at different layers. So it's all about data interconnectivity. And so, DDS is unique in the industry, in technology that's available today, in greatly streamlining that data interconnectivity.
Steven Onzo: Right. As opposed to say consumer IoT where you have a device connected to another person. With this we’re connecting things to things, and a lot of things at that.
Brett Murphy: It's very much about connecting things to things. It used to be called M to M, machine to machine. I mean, its had lots of different names over the years. This is all rolled up now into this internet of things, and in particular, our focus, industrial internet of things.
Steven Onzo: Okay, well I wouldn't be doing my job if I didn't ask you what your favorite use case is, since we'll be talking about use cases, and why you find it interesting.
Brett Murphy: Sure. So boiling it down to a single use case, I don't know if this is quite that, there's actually multiple use cases in this. It's really a case study about a particular customer's system, and that is Siemens Gamesa, as they say it in Europe. It's really their wind power group. So what they do is they deploy turbines in wind farms offshore. Now, the cell, or the center part of the turbine's the size of a Greyhound bus, and inside it there's a whole lot of compute, a whole lot of sensors, a lot of activity to keep these football field-long blades from going south. Those are expensive beasts, and they'll have like a hundred of these in a farm. So it's a big old honking system.
What I love about it is they've implemented what we call the "layered databus architecture." So they've got all these turbines, call these things smart machines, inside each turbine is a databus. There's multiple compute nodes, lots of data being shared very fast to do control of that individual turbine. Then they have another databus sitting above all of the turbines, and there, they're doing things like optimizing the output of the farm, watching for problems at that wind farm level. Then they go back to their control center, and in the control center, they have another databus.
So they have three layers of databus, and the data that's shared is peer-to-peer. They have north/south communication from edge to cloud, they have east/west communication at each layer. It's a fascinating system. And within that, they're doing control. They're doing monitoring. They're doing asset management. They have predictive maintenance. They have a whole bunch of use cases that they address in this system. It's fascinating.
Steven Onzo: That is a big system.
Brett Murphy: It's huge.
Steven Onzo: How about performance? What cases come to mind?
Brett Murphy: Performance. So one of our poster childs for performance is the NASA KSC Orion Launch Center. So the thing you see on TV where the launch center, and the people sitting in front of those displays, and the big old gorgeous display on the front wall, that's all powered by RTI Connext DDS now. And when the Orion system launches, there's a huge number of sensors on that thing just watching what's happening during a launch, and it's a burst of information, very, very, very high speed. What is it now, 500,000 messages per second during the-
Steven Onzo: Unbelievable.
Brett Murphy: ... during the burst. So it's very high performance.
Steven Onzo: It's a great example. I understand that these systems also need to deal with scale and security. How do they balance that with performance as well?
Brett Murphy: Yeah. So that's one of the unique benefits of using a standard like DDS and our implementation of it with Connext. DDS is very, very tuneable. So even the security model, for example, you can secure particular data elements, like maybe the command to shut off the valve to the launch system, shut down the fuel, that command you might want to secure very, very carefully, for obvious reasons. The temperature outside at the launch system, you could get from weather.com. Why bother to secure it?
Steven Onzo: Right.
Brett Murphy: So that sort of trade-off is doable in DDS. And then with scale, for example, they scale, again, it's back to that layered databus. So you could have a databus at the edge where you're doing extraordinarily high speed, and then you have another databus, it's essentially a logical separation of a DDS system, at the back end where you've got much lower performance requirements. And so that, breaking it up into subsystems with different data interconnectivity, data sharing needs, is how you can scale. So again, you've got all of these knobs you can tune in a DDS system to play off those different requirements.
Steven Onzo: So where is the low hanging fruit for IIoT? Where is the the wheelhouse located for this technology?
Brett Murphy: Okay, so let's step back to the overall industry right now. One of the ways that I look at the industrial IoT overall is in kind of increasing ... If you look at data interconnectivity and how much is needed, and you look at artificial intelligence or analytics, and how much is needed, you see a different set of groupings based on how complex, how much data interconnectivity, how much AI they need.
The first set of use cases I lump under the term "monitoring." Monitoring is mostly about putting a few sensors around your most expensive assets and looking at what they're doing. And the purpose is to do things like asset management, predictive maintenance, use cases like that. That's really about getting the data back to the analytics, maybe in a cloud or in a control center, and just doing ... Just doing, the analytics are actually kind of hard, right? So there's a lot of value in those. And so those are kind of the low hanging fruit. That's what most people are doing right now. They've got a factory. It's doing what it's doing. I'm going to put a few sensors in there and just start sipping data out of that. That's monitoring.
Now, oh, I've got this factory, I've got these sensors, I'm doing monitoring, I can actually add some more sensors and see what the pump in between my two expensive assets is doing. I can look at the temperature and the flow, and I can start to instrument the entire process. Now I can start to optimize the process. Now I start to need to do edge analytics, because too much data coming back. And actually it'd be really nice if I could share between the edge analytics that are happening on these different parts of the process directly, so peer-to-peer, east/west, M to M. These are where you start to see data interconnectivity needs that DDS begins to excel at. Anyway, that I call "optimization," optimizing the process
Steven Onzo: While where on that topic, would you mind explaining how the need for edge analytics, going from monitoring to optimization, instead of just monitoring these two assets, you want to add more sensors, what's the leap from monitoring to optimization? Is it too much data coming in?
Brett Murphy: Yeah. So you think about the pipe you have coming out of the factory. If you need to pipe absolutely everything that you're gathering across the entire factory, that's an enormous pipe back to the cloud. And they charge you for every byte.
It really comes down to often just physical ability, the physics, the size of the network connection, and to the cost around pushing all that data back. So there's cost advantages and just functional requirements that drive putting some of that edge analytics on the edge ... analytics on the edge. For example, do you need to send every single temperature value back to the cloud to view it? Or do you just want to show, oh, this temperature just changed over a certain amount, I'm just going to send the temperature values before and after that event back?
Steven Onzo: And this can save CPUs for the system?
Brett Murphy: It saves compute effort in the back end. It saves how much data you have to ship back, drops that significantly. So it's a variety of reasons for edge analytics. The other reason is just sheer performance. If one of your, call it an edge analytics application, sees a sudden surge in pressure at one of your expensive assets, you might want to tell the pump just upstream of that to slow down or shut off, because you don't want to blow up your gas turbine or something. I'm making that up. So that's where you want quick reaction time, edge to edge. You don't want to go all the way back to the cloud, figure that out, and send it back down. Those are obvious use cases, but a reason for edge analytics that changes from going from monitoring to optimization.
Steven Onzo: Okay, great. And you kind of touched on this just now, but what use cases will deliver the fastest ROI?
Brett Murphy: So you know when the internet first came out, the killer application was email, right? I think for the industrial internet of things, the killer app is predictive maintenance, because maintenance of expensive assets is, in itself, expensive. Typically they just say, "Well, stuff could fail after six months. We don't want that to happen, so we're just going to send somebody out every four" whether it needs it or not. Doing predictive maintenance to actually look at the condition of the system gets huge efficiency, so everybody talks about predictive maintenance. One of the techniques is condition-based monitoring. That's another term out there. That I think is the killer app that's driving the fastest ROI.
Now, the final set of use cases, on that spectrum from monitoring to optimization, final set is autonomy. And autonomy is where you basically take the human out of the loop. You're doing multiple layers of control and compute. You've got edge, you got fog in between, you've got the back end. Siemens Gamesa, right? That's a great example of one of these autonomy systems.
Steven Onzo: The wind turbines we were just talking about.
Brett Murphy: Yep, the wind turbines I was just talking about. So autonomy is huge. Let's talk about a speculative wind smart city use case. You've got a home health device that detects a heart attack, and an unconscious person in their home. It automatically calls for an ambulance, and the system alerts the hospital and sends all the data, right? The ambulance gets to the home, the EMTs, they find the person, and then they're heading to the hospital. The smart highway or transportation system turns all the lights green for the ambulance, and red increases safety, makes it faster.
They get to the hospital, all the information that's been going on with the patient as you add more sensors to them in the EMT or the ambulance is telemetered to the hospital. So they have all this data before the patient even arrives at the emergency or trauma center.
Steven Onzo: As opposed to how things are now, it's a pretty chaotic situation when you get to a hospital.
Brett Murphy: It's all siloed. You have to restart.
Steven Onzo: What's his name? What's the condition?
Brett Murphy: You have to restart each time. The EMT shows up, they have no data from the home health thing. They have to figure it all out for themselves. Then they hand it off, and they do some verbal hand-off. But the doctors have to reset and say, "Okay, what do we really have here?"
Steven Onzo: So instead of resetting, we just streamline that information straight to that doctor so they can even prepare while he's on his way.
Brett Murphy: And there isn't a person in there going, "Okay, turn the next one green at the intersection, turn the next one green in the intersection." It's all software. And that's all autonomy. I mean, it's like Facebook to email. Could we have predicted Facebook when we first put the internet together? Maybe a little bit, maybe some science fiction writers could have pulled it off.
Steven Onzo: Well, I was going to ask you how this would transform how the industry works, but it seems like autonomy is definitely that situation that you just explained.
Brett Murphy: Yes. Yep. I think autonomy is going to be the big transformation that we will see from the IIoT.
Steven Onzo: Well, RTI is involved in some of the leading consortia in IoT. Can you tell us a little bit about your involvement there and how that's enabling future use cases?
Brett Murphy: Sure. So probably the largest or most influential consortia in the industrial IoT, directly involved, is the industrial internet consortium, so the IIC. RTI is very heavily involved with the IIC, has been since its first day as a public entity. I think we were the second to join. And the big advantage of the IIC and how it's helping to shape it is, one, it's an ecosystem. It's a place where everybody ... No one company has the total solution right now. I mean, I was just talking about that smart city use case, across all these different types of systems. There's not going to be a single company that delivers all that, not for a long time. So the companies need to get together and talk about how to integrate, how to bring the different pieces that they each have to fruition. So that's the ecosystem part.
The IIC also helps with providing guidance on the technical front. So there's the industrial internet connectivity framework document, call it the IICF. That helps to define what all these different connectivity or protocol standards are, and which ones do what well. And obviously DDS is in there because we helped to author that thing, and it explains how it compares with others. So there's the technical guidance.
But enabling future use cases, the real rubber hits the road are the test beds at the IIC. And the test beds are end users getting together with a few IIC members to put together a solution for a particular use case. It could be in different verticals, it could be a different type of use case, predictive maintenance, autonomy, etc, etc. Those are really the practical demonstrations and tests, proofs of concept, for solutions for particular use cases, delivering value for a particular use case.
Steven Onzo: So clearly involvement with the consortium, the IIC, is crucial to progression in this industry.
Brett Murphy: Yes, because otherwise we'd have wild, wild west chaos. Who knows.
Steven Onzo: Right. It needs some organization.
Brett Murphy: Right. Now, it really does accelerate the industrial IoT in its entirety. And that's its purpose.
Steven Onzo: And kind of paves the way for how this industry will look, since this hasn't been solidified yet.
Brett Murphy: Nope.
Steven Onzo: All right.
Brett Murphy: We're at the first steps of the IoT, and who knows, what's the IIoT Facebook going to be?
Steven Onzo: Right.
Brett Murphy: Equivalent.
Steven Onzo: Excellent. Well, thanks, Brett, for sitting down today and talking with us. If the listeners wanted to learn more, do you have a recommendation on resources where they can learn more?
Brett Murphy: Sure. I mean, the IIC has a whole bunch of resources that provide kind of general guidance. If you're interested in seeing how that guidance can be applied, how the IIoT is evolving, I actually suggest going to rti.com and grabbing the ebook, The Rise of the Robot Overlords, provocatively named, but it's really about, what's the IIoT industry, how is it evolving? How is it going to evolve? It's an ebook, but then if you want to get even more detail, there's six coming up soon ... seven webinars that Stan Schneider, RTI CEO, has put together. And so, that will go into even more detail.
Steven Onzo: Excellent. Well, Brett, thanks again for joining us and allowing us to pick your brain. It's been great. And to all the listeners tuning in, we'll see you next time. Thank you very much.
Brett Murphy: Thank you.