Huge Challenges With Billions Of Things

Advancements in technology are about to open new doors to communications. That’s both good and bad.


Communication is poised in the next couple of years to cross a line between humans and things—things talking directly to other things as well as to people—setting in motion a series of technological, social and legal issues that will take years or decades to resolve.

On one hand, this is made possible by leaps in processing performance and power management in mobile devices. In his keynote speech at ARM TechCon 2013 last week, ARM CEO Simon Segars noted that we are entering the era of mobility where it’s not just about access to the Internet—it’s about access to the mobile Internet.

“There is a growing network where the phone becomes the interacting devices to a network of other things,” Segars said. “But with that the consumption of data is going up. Global data grew 50% this year to 900 petabytes, and Cisco predicts by 2017 mobile data will increase by a factor of 10 to 8 zetabytes. The network today can’t absorb that much data.”

Data overload is just one problem that has to be contended with for the Internet of Things. Across every industry, including semiconductors, the IoT is almost universally recognized as a huge opportunity. It also poses a long list of challenges that crosses all boundaries.

Technology first
Just getting the IoT to work in the first place is a big challenge. While the general perception is that many sensors are relatively simple, relying on older process technologies with an emphasis on cost, the reality is that intelligence has to be built in at every phase of the connected world. Not all data should be transmitted, and being able to figure out what, when and how far that data moves is more than a simple hierarchy of permissions in software.

“As long as there is not much need for filtering, it’s relatively straightforward,” said Frank Schirrmeister, group director of product marketing at Cadence. “So the sensors may be relatively simple, but there is more computing on the sides of the sensors. That could mean high-end drivers. You also might have hubs in between that accumulate information.”

But he said the big question for many people and businesses is where exactly the IoT connects to their world. Is it inside the device, the building, or something further away? “Not all applications need to go into the IoT,” he said.

So what will work where? That question is being asked a lot these days.

“The IoT is currently an extremely fragmented market, so there is not a “one-size-fits-all” answer,” said James Wu, director of strategic marketing at Synopsys. “We have seen designs implemented in 90nm/55nm embedded flash processes or 40nm/28nm logic processes. It all depends upon which particular sub-segment suppliers are focused on. For example, an MCU for a parking meter or vending machine can be implemented in 90nm or above technologies. These meters are using the 2G Wireless devices to connect to the Internet, which will be designed in 40nm with less than $10 BOM (for the wireless modules). On the other hand, the future intelligent cars with driving assistance or collision avoidance could be designed in 28nm process technology or beyond.”

Data from multiple sensors also can be combined to achieve more functionality, which is the basis of sensor fusion or smart sensors, said Wu. That will require a far higher level of processing than the 4- and 8-bit MCUs in use today, though, and it may require rethinking what processing is done by the application processor and what is done by the MCU.

Also at issue is how things will communicate with each other. Some of the communications are wireless, but some of it also will be connected by wires.

“It’s still unclear what standards will be needed,” said Eran Briman, vice president of marketing at CEVA. “Some of that may be geography and application dependent.”

The good news is that some of it can be handled by the increasingly heterogeneous network (HetNet) technology, which provides a lot of flexibility in terms of data types, numbers of users and the mix of users. But how much volume this sophisticated network ultimately can support is anyone’s guess, and it may vary by location and the type of data being transmitted.

Power down
One of the most daunting aspects of the IoT is that many of the things are always on, so power becomes a critical problem to solve. That doesn’t mean things are always transmitting, but they do have to record changes in the environment they’re measuring. In the case of a pacemaker, that might be a second or less. In the case of a temperature gauge, that might be measured in minutes. And in the case of a camera, it might be continuous.

There is no battery on the planet powerful enough to send and receive data all the time without a recharge, which has put a big emphasis on energy harvesting technology.

“Even if you run at exceptionally low power, in many cases you’re not going to want to have to change batteries,” said Bernard Murphy, CTO at Atrenta. “So you have to think about energy harvesting and microwatt power, which means advanced communication technology and very advanced sensors. This technology is not trivial. There are solutions out there right now for this. They may not be widely deployed until devices take off, but they do exist. The other choice is RFID tags, but you have to be really close for those to work.”

Five years ago, most of this was more promise than reality, but this kind of technology actually is being manufactured today.

“We’re already seeing this for microcontrollers and the Internet of Things,” said Srinivas Nori, director of SoC innovation at GlobalFoundries. “The real challenge is that this stuff has to be designed from the ground up. With harvesting you get unlimited power but finite current, so you can run fast and then sleep or run slow all the time.”

Standards, or lack thereof
Another big problem is standards and consistency. Dennis Brophy, director of strategic business development at Mentor Graphics and corporate advisory group chair for the IEEE Standards Association, said he held meetings in multiple countries where each of them claims to have created the IoT.

“There is a global standards underpinning to all of this,” said Brophy. “We want things to work seamlessly with other things everywhere. There are a lot of different groups working on the IoT and so many people and countries that claim to be the father of the IoT. What that means is we can’t depend on countries coming together on this.”

From a communications standpoint, that’s not necessarily a good thing. And from a privacy and security standpoint, it may even be worse.

“Standards enable the possible,” said Brophy. “It’s more difficult to control the unwanted.”

Leave a Reply

(Note: This name will be displayed publicly)