Skip to main content
Exploring what’s next in tech – Insights, information, and ideas for today’s IT and business leaders

The IoT: Where real-time insights drive real-time action

In this episode of Technology Untangled, experts take a look at the hottest IoT trends, including analytics at the edge.

When people talk about the Internet of Things, what are they really referring to? According to Ian Henderson, chief technologist at Hewlett Packard Enterprise, the IoT is simply "distributed technologies at a massive scale," encompassing an infinite number of devices creating data.

Beyond the nuts and bolts, "the real value of the interconnected thing is the data, and that needs to be shared and visible across the whole organization," Henderson says. "So that's the challenge: not just generating more and more data but getting value and visibility of that value."

That's why analytics at the IoT edge is where the future lies, says Nitu Kaushal, HPE's general manager of compute, edge, IoT, and digital services in the U.K. and Ireland. For example, she notes, the pandemic has increased demand for IoT analytics in airports and train stations, where data is captured from connected devices to measure people's temperatures, assess the safety of crowds, and bolster control points within certain areas.

This episode of Technology Untangled takes an in-depth look at how the IoT and edge analytics are producing real-time insights to support real-time action across a wide range of use cases, from preventing coastal erosion and streamlining shipping logistics to improving customer experience and increasing operational efficiencies. Also learn the latest on remote monitoring at the IoT edge from Chris Roberts, head of artificial intelligence and cloud at the Goonhilly Earth Station.


Excerpts from the podcast, hosted by Michael Bird, follow:

Michael Bird: What exactly constitutes our definition of IoT today? And when does it become really useful?

Ian Henderson: One of the customers that I've been talking to recently is a water utility, and they have what they call a SCADA network, so an industrial control network. And they might describe that as Internet of Things, but the reality is it's very much a status-based network with preprogrammed logic. In crude terms, it's a bit like "if this, then that"―when this happens, open this valve; when you see that happen, close this valve. And they can see if a valve is open, and they get those updates maybe every few minutes, not in real time. What they can't see is that the valve is open but the water isn't flowing because there's a lot of rags and rubbish blocking the pipe. So what they really want to do now is add additional sensor data so that they're looking to get real-time data of what's actually happening.

And that's when it really becomes IoT for me. It's just not, "I can see what lights turn on or not." It's how much power is the light using? How do I add a few more sensors to help me understand what's going on? Not just that the pump is running, but actually what's the water pressure on either side of the pump. If I can understand the water pressure on either side of the pump, I can tell probably how fast the water is flowing, which gives us a system of insight. So I'll have a better view of what's happening at the edge.

A system of action

The next level beyond that then is where I turn that into a system of action, right? We're going to connect some additional sensors, we'll almost wiretap some of the existing network, and we will take that data and allow visualization to understand how things are operating and then look at the way I operate today and use that data and modify the way I operate. But the point that we like to get to for that next level of efficiency is where it's closed loop, where I'm taking the extra sensor data and I'm feeding back to an actuator. So I'm looking at how our pump is running and maybe changing the speed that it operates.

Bird: So it's not just a case of connectivity equals good. Benefiting from the IoT is all about the value of data.

Nitu Kaushal: The automotive industry we know is moving towards software-heavy cars. This allows cars to run updates over the air and have a much better driver experience within the car. So connecting the cars and then connecting the manufacturing plants provides a lot more clarity and better product into the market for the automotive industry.

Henderson: We've worked with the Mercedes Formula One teams, one of the organizations that we work with, and you could almost look at it that they've invented IoT, right?

Those vehicles have typically around 300 sensors on them, and they're looking at every single piece of that data. And that's a sort of bleeding-edge, really good example of how connected that you can get, that they're actually predicting how many more laps they can get out of a set of tires based on all sorts of data.

I think that's leading edge showing how you can exploit masses of data in real time and then change the way that you operate. The way that I see those guys taking that data and understanding exactly what's going on with the car and changing their strategy based on that in real time is probably the sort of pinnacle of how you get value from that data.

Bird: In the automotive industry, edge devices taking in massive amounts of data allow real-time tweaks to the car and its performance.

As we mentioned on our previous episode on edge, there is a pendulum swing around to where the data is processed. Compute at the edge is becoming more and more common as "things" become more powerful and the datasets grow exponentially larger.

Real-time insights = real-time action

Kaushal: Where IoT is about connecting assets and collecting data from them, edge becomes the ability to process that information much closer to where the asset is and then be able to provide near real-time action-led insights back into the business. Live video streaming all day long is capturing a lot of information, and if all these video streaming files are then processed in a central cloud, we introduce additional time needed to be able to analyze that data. That predictability that a business wants to introduce becomes a lot harder and slower. So, if time is of the essence, then some of these applications need to be processed at the edge.

Bird: Being able to process data at the edge means the IoT can provide innovative ways to manage and monitor far-flung operations in remote places, constantly feeding information into applications and data stores.

Kaushal: In the space of agritech, what's been really interesting here is whilst we are in remote spaces, connecting up some of these areas has always been a great difficulty. But with the rise of 5G and also low-power wide-area network narrowband IoT, the ability to connect these remote spaces and also connect low-value assets has become a lot easier. With the ability of drones, we're now able to fertilize land remotely and have much better control and information of how some of these crops are progressing over time.

Bird: To hear a little more about remote monitoring at the edge, we called up Chris Roberts, the head of artificial intelligence and cloud at the Goonhilly Earth Station.

Chris Roberts: Goonhilly is effectively a telecommunications hub.

Effectively, all of the satellites or birds that are flying around the sky collating information, or if information is bouncing off them, that will either be emitted from, caught by, or a bit of both by Goonhilly. Then the information will be sent, let's say, from Goonhilly and transmit via satellite to a location somewhere else on the globe.

Fifty, 60 years ago, when Goonhilly came into life, all of the U.K. satellite traffic pretty much went through Goonhilly. So think of it as a sort of central hub, if you like, for all of the satellite comms in the U.K.

Goonhilly use case: Space

Bird: Goonhilly sits on a 163-acre site, and looking at it, it looks super-futuristic with all these big dishes, or aerials, as Chris calls them. And it's based near Land's End in sunny Cornwall, the westernmost point of England. And Goonhilly was one of three earth stations in the world involved in the first transatlantic TV transmission in 1962. It beamed the moon landings to millions of viewers in 1969.

Now, like many organizations, they're pretty interested in what's happening at the edge. The difference is Goonhilly's edge is in space.

Roberts: There's two primary functions, which are deep space and near space. Deep space is all about a commercial deep space network. So, if you think about NASA, ISA, all these space agencies around the globe, some of the assets that they were using for deep space communications are decades old, and the idea really is to provide a better, more efficient network that can support the space transmission and communication requirements of lots of different countries.

So near space is what we call anything that isn't deep space. This is the low Earth orbit and geostationary stuff. You've heard of Elon Musk's company SpaceX, Blue Origin, Jeff Bezos, and this is where they are building out, if you like, a space Internet. With deep space, your target market is space agency, so there's a limited amount of those on the planet. So it's still quite large dishes.

And the low Earth orbit satellites whizz around the earth at very high speeds, so they need much more nimble, smaller antennas that can track these. We have both the deep space antenna and the low Earth orbit antenna at Goonhilly, and then we also use geostationary antennas as well. And these are all sitting at different orbits in the earth.

Bird: There are lots of reasons why we'd want to take readings and monitor the earth from space: weather and navigation, earth observation, telemetry and tracking, and even deep space communications. And all of this edge data is inspiring a host of previously unexplored use cases.

Low cost, big potential

Roberts: Edge, particularly for us, does lend itself towards things such as metrological, so weather, shipping―anything that requires the data to be acted upon quite quickly or where people want to be able to process that data in close proximity to the antenna.

Because of our proximity to the subsea cables, at Goonhilly in particular, we sit on, we're almost like a springboard for, 38 different countries. Data can come in from all over the world, and what we can do is take that and process that and potentially send that back out again, without even touching the kind of core networks―which has got an extraordinary commercial application, if you think about it, reducing the cost of that data.

I mean with earth data, although it's a very sort of catchall term, the applications are extraordinary. Look at coastal erosion and flooding. So, if you think of insurance companies, one of the biggest challenges for consumers at the moment when buying their insurance … you might be half a mile away from another guy who is living in a basin that gets flooded every year and you're absolutely fine, thank you, but they can't be as accurate in [calculating] premiums.

Using earth observation technology to sort of look at the patterns of flooding, you can be much more accurate about that site to build a real picture. That's collating data and storing that at the edge.

Bird: What makes Goonhilly so exciting is the way this edge data can be used to make decisions and potentially change industries.

Reducing environmental impact of moving cargo

Roberts: I think that's going to happen as the data volumes increase and the possibility of autonomous communication between things like self-piloting ships. There's a vast amount of money going into the digital marine marketplace in order to improve things like the environmental cost of moving tons of stuff around. So all the stuff that we order from China or wherever has an environmental cost or a carbon cost.

So you can do things like track the telemetry of ships. Then, if a ship's got space, if a ship's low in the water, it's carrying a lot of cargo. If it's high in the water, it's probably not carrying anything. We're already working with a couple of companies who are looking at tracking satellite data, tracking telemetry of ships, for a number of different reasons.

But essentially you could say, right, I'll look at all of the ships around the world, look at what their unladen status is and their laden status. And you might say, well, we reckon that you could fit 40 percent more cargo in that ship without it being overladen. So there's lots of different use cases for this, and if you can start to almost point vessels or devices and make them more efficient, you do that by processing that at the edge.

Bird: Now, if coastal erosion or streamlining shipping operations isn't really your wheelhouse, don't worry - as Ian explains, the IoT is all about driving efficiency in just about any industry.

Efficiency, customer experience, real-time analytics

Henderson: A lot of the time, working with the larger organizations, there tends to be three things they're trying to do, right?

One is producing operational costs. So a lot of that is looking at overall equipment efficiency, understanding if I've got machinery running in the location, and is it running as efficiently as it could. Can I compare that to a similar facility that I have somewhere else in the world and understand why one of them is more efficient than the other?

The bigger one probably is how do you improve the customer or the employee experience? That means things like location- based services, so the ability to guide a customer or an employee around a facility using Bluetooth, the ability to understand how many devices are in a location and how they're working―those sorts of use cases.


And then as we go into the more extreme, the analytics at the edge, which is certainly where we're spending a lot of time looking at what those use cases are. So video analytics, particularly in markets like retail: How do they give a different experience when a customer comes to the store? And a lot of that is around frictionless―the Amazon Go example [is] where I can walk into a store, it recognizes me as I walk in, I walk around and pick up my items and walk out, and it automatically builds my account.

How do I do that whilst not risking increased shrinkage, as they call it, so theft or mis-scanning of components.

Bird: Analytics at the edge is the hot topic in IoT at the moment, and Nitu thinks we're going to be seeing a whole lot more of it in the post-COVID world.

Use cases post-COVID

Kaushal: We are seeing other emerging use cases for IoT, and some of these have been driven by the pandemic itself. So [that means] thinking about existing CCTV footage that buildings and businesses have been capturing for years, thinking about how that could then be used as a safety mechanism in airports and in train stations―that live streaming video, measuring people's temperatures, sharing that video back to a control point that will then analyze the safety of the crowds and be able to use that information to encourage better safety procedures and control points within areas. And this really shows that a lot of data is going to need to be processed at the edge because you're going to require rapid decisions.

Bird: Organizations approaching the IoT will need to consider how they deal with data, both from an analytical and organizational point of view.

Henderson: A good example: A European automotive manufacturer had different simulations of data being run by the powertrain team, the chassis team, and the noise vibration and harshness team. And when we looked at the data, some of the simulations that were being run by the noise operation and harshness team had already been run by the chassis or the powertrain team earlier on in the process.

So the different teams described the data differently, and they don't share it. And this was actually one of the questions from the motorsport team that we worked with where they said, "We know the answers to everything, but we don't know what we don't know."

But I think that's the real value here of the interconnected thing is the value of the data, and that needs to be shared and visible across the whole organization. So that's the challenge: not just generating more and more data but getting value and visibility of that value.

Proof-of-concept purgatory

Kaushal: One way [businesses are] looking at near real-time data [is] that it can help them drive some near real-time business actions. The other way is looking at things in a more data-at-rest perspective, from a data lake, a big data perspective, using these insights to build trends over a longer period of time that can help businesses really understand how they are operating but also what their customers need.

Henderson: One of the challenges I see in IoT is what we call proof-of-concept purgatory. How do we build the business case for these things?

Lots of people have very smart ideas about what they could do. They'll take a piece of relatively low-cost hardware like Arduino or Raspberry PI and show a simple model of how they can get the IoT data that they want, but it never gets to that point of scaling out to be real-life production use.

It's really about the value of data through the full lifecycle of the product, taking data from the connected products and feeding that back into the development and continuous improvement of the design. You can look at it with Tesla right now: The vehicle you buy is not a finished product; it can change and be upgraded over time.

But to build some of those solutions, it's very hard to get the business case over the line, right? Because you're saying I want a really big investment to do all of this huge project and I've not really proven that the business case is going to return the investment that I make.

So the way that I work with customers is trying to define, where's that nirvana that we want to get to? But how do I break that down into a series of small sprints? How do I do a six- or eight-week project which shows I can return the data and the value that I want, with a second step of then saying, right now, we're going to scale it out across the next six or eight weeks to another 10 locations. And the business case gets built stronger and stronger as you move through it.

Bird: Although the IoT has a lot of potential in many ways, it's still a maturing innovation.

Dawn of a new era

Kaushal: Machine to machine has been around for over 10 years already, but if we look at the grand scheme of things, we probably haven't connected even 5 percent of all the devices that could possibly connect in the world. So there's plenty of space to move into. And new technologies like 5G and narrowband IoT are going to enable the connections of that remaining 95 percent of devices.

So, in that sense, you could think that it's still a new industry and new technology, but the understanding of IoT has definitely matured. And the ability to integrate IoT into your home is also an ongoing trend in urban areas.

Bird: Industry has done a lot of the legwork in the IoT story, and they're pinning their hopes on―buzzword alerts―Industry 4.0, known as, potentially, the next Industrial Revolution.

Henderson: I think we've got a long way to go. In the next 10 years, I think IoT is going to be amazing.

You know, if I look into manufacturing, they haven't really started yet. So you'll hear this term, Industry 4.0. We went from steam to electricity and to automation; these were the first three. And the way we described the fourth Industrial Revolution is moving from automation to autonomous. Autonomous is that closed loop of data management.

Automation, which we have today, will do the same thing consistently. If it drifts out a tolerance and goes wrong, it will just produce something consistently wrong. I used to get companies that would say, "We've got ISO 11,000, is it? You know, we're a quality company."

Well, it doesn't mean you're making a quality product. It just means you're making it the same way every time and you can prove it.

So the ability to move from automation to autonomous development and self-tuning is what Industry 3.0 is about. The challenge at the moment is we haven't even really started getting the value from the data that we can collect.

So I think it's going to be really fascinating.

Key IoT considerations

Bird: So what do organizations need to develop in order to use the IoT, make the most of the data they're producing, and aim for that closed-loop management?

Henderson: In a lot of these spaces there's a healthy skepticism and concern about the security risk, and that's good because there is definitely a risk that needs to be considered and addressed as you go into it.

I think one of the challenges as we roll these out is how do I manage all of those devices? It's not too difficult to take a Raspberry PI or an Nvidia Jetson and show how I can do some really powerful analytics at the edge of the network, but how do I do the lifecycle management of it to do firmer updates, application updates? And those are challenges that we've learned in the data center over the last 20 years when we sell a high-performance compute system with literally thousands or tens of thousands of calls ... the ability to understand the health and manage that.

That's where we're investing a lot of time, looking at how do I manage the security and lifecycle of firmware and applications across potentially thousands or tens of thousands of connected endpoints all around the world. And I think that's the level of maturity that we need to get to.

Bird: A robust security architecture for devices and edge networks is absolutely paramount for organizations, and you can find out more about that in our previous episode on edge. For Nitu, the proliferation of 5G is going to be vital to help smaller organizations advance too.

Kaushal: The network providers play a really strong role in firstly rolling out new technology as a backbone to enable more connected devices. But then also how you can access some of these frequencies a lot more easily to drive innovation locally within different countries also becomes important. So we've seen in the U.S., for example, in September 2019, some of the operators have made frequencies for free so that businesses can drive net new innovation. And this is helping smaller businesses accelerate.

Bird: There's already way more things connected to the Internet than people. So what on earth is the IoT going to look like in 10 years' time?

Henderson: If I count myself back 10 years and think of what it was ... and I was a hobbyist playing with home automation. I was using the thing called X 10, which signals over power lines to turn lights on and off and have some logic into the way that my house worked. And it was very much a bleeding-edge, hobbyist technology ... not very reliable, quite expensive.

If I look at the house now, you know, everybody's got Hue lightbulbs, hive heating―it's mainstream. So I think if we look forward 10 years, these things will just be ubiquitous. We will just assume this visibility of data and the ability to interact with our world in that digital way is just natural.

Bird: You've been listening to Technology Untangled, and a big thanks to today's guests: Ian Henderson, Nitu Kaushal, and Chris Roberts.

And you can, of course, find out about today's episode in the show notes and be sure to hit subscribe in your podcast app and join us next time when we'll be charting the murky waters of digital ethics: artificial intelligence, big data, and the challenge of our generation.

Today's episode was written and produced by Isobel Pollard and hosted by me, Michael Bird, with sound design and editing by Alex Bennett and production support from Harry Morton and Alex Podmore. Technology Untangled is a Lower Street Production for Hewlett Packard Enterprise in the U.K. and Ireland. Thank you so much for tuning in, and we'll see you next time.

Listen to other episodes:

What AI means for you and your business―now and in the future

As-a-Service: What it is and how it's changing the face of IT and business

Sustainable tech: Good for the planet, good for business

Hyperconverged infrastructure: Speed, flexibility, performance in a box

Containers: Big innovation in a small package

The edge: Data wherever you need to be

This article/content was written by the individual writer identified and does not necessarily reflect the view of Hewlett Packard Enterprise Company.