What Exactly Is The IoT?

Beyond the hype: The vision, what’s missing, and what else is needed to make the Internet of Things fulfill its promise.


The Internet of Things has drawn its share of attention. It has been overhyped, derided, and in some cases dismissed as the Internet of Silos. But the idea persists, and it’s catching on.

So how much substance is there behind the hype? Best guess: Quite a bit.

“We do see this as being a fast-growing market,” said Jim Feldhan, president of Semico Research. “The number of connected devices over the next five years will exceed the installed base of mobile devices. There is a hockey stick coming, and there are a number of different avenues to get into IoT. One of them is through lighting and LEDs, especially in the commercial space. One of the opportunities is using power over Ethernet. If you’re in a commercial building there’s a mass of wiring. That can all be eliminated using your Ethernet network and connecting all your lighting. That helps you consolidate your network. Once you have your lighting through Ethernet, it’s easy to add surveillance, environmental sensing, track your elevators and incorporate your fire and safety. That all connects to your network and into the cloud.”

Another big avenue for connectedness is the home. “Over the next 10 years we see 23 billion residential appliances that will be connected to the Internet worldwide. From a semiconductor standpoint, that’s going to equate to about 160 billion semiconductor devices. Compare that to the total IoT market, and we think the residential market is about 20% of the total. So it’s a substantial market over the next 10 years and beyond.”

Feldhan said one of the problems so far is that companies are using existing technology and trying to fit it into new markets, whether it’s wearables, smart homes, industrial, enterprise or even smart cities. “Nobody has come up with a solution that knits everything together yet and creates a barrier to entry,” he said. But there is progress being made. He cited an application being used in Nice, France; Barcelona, Spain, and on the campus of Harvard University, where sensors are being put in garbage cans and trash removal only occurs when the cans are full. He said that saves time, energy and money.

What’s missing?
Still, this is hardly revolutionary stuff. Terry O’Shea, distinguished technologist at Hewlett-Packard, said this used to be described as telemetry. Then it became wireless sensor networks. But what always has been left out of the big picture is the integration of all the pieces.

“When you start to break it down, even in large market segments, you can break out into large trees everywhere,” said O’Shea. “If you look at energy, there are 8 or 10 things there. In fitness, there’s a plethora of things, from smart gyms to fitness devices you wear. It goes on and on. The issue is that it’s a huge market. Where do you start off? You really need an agile set of building blocks so you can grab pieces from sensors, from embedded processors, from code, put them together on a similar security basis, put them on a device and send them up to a cloud server and start charging money. The sooner I can put those pieces together and enter into every branch of this market segment, the more my cloud grows and the more value that’s there.”

He said that edge devices such as wearables or motion controllers are only the beginning. “If you look down there is a whole ecosystem starting with silicon. We can package that together onto a board that goes into a case. There’s industrial design. That connects to a client device. That client communicates with something, which is usually a router in the home, which goes through some ISP. Then you get into service enabling, service provisioning for adding services to this. It turns out that often we don’t design these systems for the end user. We design them for the customer, and the customer isn’t always the end user. Oftentimes that’s a reseller or another customer. From the silicon providers, I need an integrated set of tools that allows me to go all the way across the value chain, from the very edge device all the way up to the cloud services. That set of tools is so similar that I don’t need to hire a firmware guy on this side, a specialist in HTML on the client side, and I don’t need to hire a bunch of IT guys. If that was continuous, I’d have an easy time integrating all these pieces together.”

What’s interesting about the IoT is that most people involved with it have examples of how this can work. They’re usually in specific markets, but they’re intriguing enough to capture your interest.

O’Shea’s involves a license plate card reader in Brazil that was set up by Intel. It reads 428-bit keys from 100 feet away on cars moving 100 MPH, but it also can operate for five years on a coin-cell battery. The system is in place for 43 million cars in Brazil, and it’s tied in with the national gas company, Petrobras, which displays information about cars and driving when drivers fill up with gas.

“If you look at what connects power, the sensors are really what grabs the power,” he said. “If you start integrating sensors and ADCs together, they’re consuming the most power because they’re always on. The processor, for the most part, on these edge devices, is usually asleep. The second thing that takes up the power is the RF and the communication with that edge device to its client or aggregator. We can produce a large volume of these devices, but who’s going to change the batteries on them? If you have 7 billion devices, Eveready doesn’t make that many batteries.”

Making things—and people—work together
As the saying goes, it takes a village to raise a child. It will take a global system of villages with lots of standards to make the IoT work properly.

This may sound impossible, but the Internet and global smart phone coverage have been built largely using the same approach. VK Raman, vice president of foundry engineering at Broadcom, said that by 2019 there will be an estimated 5.6 billion smart phones on the planet.

“That is going to trigger about 30 billion Internet of Things devices,” Raman said. “The requirements are going to be slightly different, but this is going to go incandescent.”

Within this system—connected homes, health monitoring, connected cars, smarter industrial machinery—everything will have an IP address. “Security is going to very, very important,” he said. “So will scalability. When you talk about 30 billion devices, any software, any interface, is important. And cloud technology needs to be very easy to obtain and reliable.”

All of these devices will require rapid connectivity, and it needs to swap easily from localized Bluetooth or WiFi in the home to broader connectivity when people are on the move. And for this there need to be standards.

The building blocks are simple, but the whole ecosystem is really, really complex,” Raman said. “If a driver gets sleepy, a sensor can detect that and alert him. This will change the life of the human being, and this is happening in front of us.”

Connected at any process node
What’s also intriguing about the IoT is that it is a system of devices that are process-agnostic, all working together in unusual ways.

John Koeter, vice president of marketing in the solutions group at Synopsys, said that machine to machine communication will probably happen at 90nm, while wearables will use 65/55nm technology, and high-end smart watches will begin at 28nm.

Being able to connect at any node opens some interesting doors for devices. For one thing, it allows chipmakers to use the most cost-effective technology for their application.

“This market is all going to be about cost, so having very aggressive constraints on your design in terms of the number of metal layers in the most cost-effective packages will be really important,” said Koeter. “It used to be that the processor element in a sensor was 8-bit, but we’re now getting data coming in from many places and processed in a sensor hub or sensor fusion. So we need to move from 8-bit to 32-bit microprocessor systems.”

Examples include smart blankets and connected slippers. Koeter said that with smart blankets you can swaddle your infant and still monitor their heart rate and breathing, while smart slippers use accelerometers to gauge whether the elderly have taken a fall. Both send alarms to a smart phone or other monitoring device.

Recipe for mass deployment
But to really get to mass production and integration, the key in the IoT will be the fundamental driver for the semiconductor industry—economies of scale driven by repeatable processes.

This is where collaborative innovation is required.

“If you look at the drivers for the technology development, it used to be the technology itself,” said Subramani Kengeri, vice president of design solutions at GlobalFoundries. “But if you look at what’s happening today, it’s the applications that are driving the technology. For the foundry, or the manufacturing, we have to start looking at the applications to understand the requirements and the future needs. That is what is needed, and that is what is starting to drive the technology definition today. Time to market has really shortened. Today, the product development cycle is short, but we also want better technologies on a much shorter cycle time. All of that adds a lot of complexity. It’s possible to get there through collaborative innovation. It requires standards. Otherwise, if everyone wants to devise their own and there are no standards, they will remain things, not the Internet of Things.”

Kengeri noted that there are 3,000 startups working on wearables, sensors and 3D displays. “These put tremendous pressure on technology requirements, because on top of this you want really low cost. One thing we do is collaborate.”