IoT Predictions for 2018

In 2016, the Internet of Things (IoT) reached enough popularity to earn that year the moniker “Year of the Smart Device”. In 2017, the number of IoT devices surpassed the number of humans on Earth. In 2018, one can’t help but wonder: what exciting IoT developments are coming this year?

To help answer that question, we spoke with several industry insiders for their take on the IoT trends to watch out for this year:

• Miguel Castillo, former CEO and founder of Carriots, now VP of Global Business Development for IoT at Altair
• Neil Hamilton, VP of Business Development at Thingstream
• Julien Calviac, IoT Cross-Industry Solutions senior director, Dassault Systèmes

IoT Networks in 2018

There’s been a recent uptick of interest in low-power wide area networks (LPWANs), a breed of connectivity protocol tailored for the IoT, and it’s not surprising that this interest will continue through 2018.

“I'm starting to see requests from customers using LPWANs like Sigfox or LoRa,” said Altair’s Castillo. “I see a lot of companies interested in LoRa technology because it’s more versatile.”

As the name suggests, LPWANs serve the IoT in two ways: they use very little power, which is critical for IoT devices meant to last for several years on one battery, and they can cover a wide area, which means IoT devices can be effectively utilized out in the field. This latter point is especially important for Industrial IoT (IIoT) applications, such as agricultural sensors spread out across a farm.

“The existing challenge is to bridge the ‘air gap’ between these devices and the cloud,” said Thingstream’s Hamilton. “Achieving ubiquitous connectivity—and security—is therefore a major obstacle in fully enabling a worldwide network of IIoT devices, which will in turn allow the wider IoT spectrum to grow.”

There are two flavors of LPWANs working to bridge the air gap that Hamilton describes: cellular LPWANs like NarrowBand IoT (NB-IoT) and Long Term Evolution for Machines (LTE-M), which use cellular frequencies; and unlicensed LPWANs like LoRa, Sigfox, and RPMA (Random Phase Multiple Access), which use publicly available Industrial, Scientific, and Medical (ISM) radio bands.

Of the unlicensed LPWANs, Castillo has the most faith in LoRa. “From those three, I think LoRa is probably better positioned than the other two. Because from a technological point of view, LoRa is capable of sending more data, so you have a larger bandwidth. And you don't really depend on an operator. Sigfox is an interesting technology, but it has to create a lot more momentum or critical mass in order to support itself,” he said.

Hamilton endorses a different type of network for IIoT applications: Thingstream’s bread-and-butter, Unstructured Supplementary Service Data (USSD) messaging over mobile GSM networks. “Using a cellular or low-power wide area network doesn’t guarantee ubiquity of connectivity [to bridge the air gap] or ensure secure delivery of all connections, due to the vulnerability of Internet-connected devices,” Hamilton explained.

The Rise of Predictive Analytics

2018 may be the year when we start to more fully cash in on the promise of IoT data.

“In this year ahead, the words ‘predictive analytics’ are definitely going to become mainly dominant,” predicts Castillo. “We're in a stage where all of these companies have already started collecting data from the products, and obviously the first thing that you want to do is monitor your products. And the second stage is controlling them. But the next step is trying to optimize or predict the data that is coming from those devices. It could be for predictive maintenance, or it could be for predictive consumption, but getting to predictive analytics is a little more complicated than jumping from pure monitoring to control.”

As it stands, IoT sensors are generating massive quantities of data, but unlocking the full value of that data remains a challenge. One of the biggest hurdles to unlocking that value is simply the sheer amount of computing power needed to sift through it all, according to Castillo.

“In the years ahead, what we want to explore is how we can leverage high performance computing to process even more data,” Castillo said. “For example, we have a customer implementing IoT in manufacturing for the automotive industry. For every piece that they manufacture, they collect 1,000 variables for that part. But right now, they are only capable of calculating predictive analytics on 250 variables—just one-quarter of the total number of parameters they're collecting.”

Virtual Product Development

According to Calviac of Dassault Systèmes, another trend to watch out for is the increasing complexity of the software in IoT devices.

“Roughly every seven years, the number of lines of code in any given product multiplies by a factor of 10,” Calviac claimed. “This growth rate is showing no signs of slowing down, and may increase as our appetite for devices with embedded Internet functionality increases.”

What does this increased complexity mean for IoT developers? In order to manage the tsunami of IoT-driven software to come, businesses must increasingly turn to virtual product development to help refine their IoT offerings.

“The rise of virtual product development will be the key for businesses to increase complexity while reducing costs, especially as businesses add unfamiliar functionality via complex software to their products,” said Calviac. “The frequency of product updates will also increase, as feedback from IoT-enabled devices will allow teams of product managers and software engineers to analyze mountains of user-generated data, looking for ways to continuously improve their products.

“But businesses can’t afford any more budget or time to market, creating numerous iterative physical prototype variants,” Calviac continued. “To ensure interoperability across increasingly complex systems, software functions and mechanical capabilities will need to be validated and tested using virtual replicas of products. But just as importantly as testing core product functionality, the high-tech industry will place increased value on testing the final ‘real-world‘ customer experience using advanced product simulations—all before any physical materials have been manufactured. It’s the only way for businesses to keep up with competitors and consumer expectations without blowing their budgets.”

On the IIoT side, simulations of processes will be just as valuable as IoT product simulations, according to Castillo.

“As we proceed into high performance computing, we're going to be able to process more variables, but also process different scenarios where you could imagine variations on those variables,” he said, “to find other possible manufacturing scenarios or operational parameters of the machine that you don't have in reality, but could explore in simulation, to try to optimize a machine or performance.”

Tip of the IIoT Iceberg

Lastly, expect to see some IIoT investments starting to pay off for early adopters in 2018. Whereas there are a lot of consumer IoT products on the market, the IIoT has been lagging behind by about three or four years, in Castillo’s opinion.

“Now, I see leaders in their industries starting to implement these [IIoT] technologies after some proof-of-concept phases, and starting the return of their investments. Those companies are really starting to convert the raw data into useful information, and turn that useful data into prescriptive actions into how a process should be changed or how a machine should be reconfigured.Not all the companies are capable of doing that process…. We're definitely at the tip of the iceberg,” Castillo said.