The industrial internet of things (IIoT) -- using IoT technologies in manufacturing -- is beginning to take off as more enterprises employ industrial IoT platforms for such tasks as performing preventive maintenance, tracking assets in factories and improving overall plant operations.
Yet, the use of industrial IoT platforms in manufacturing is being held back by a lack of interoperability between devices and machines that use different protocols and have different architectures.
To fully realize the advantages of IIoT technologies, interoperability standards and common architectures are needed to connect these smart devices and machines. But developing such standards is often difficult because stakeholders, including vendors and industry associations, usually endorse different approaches.
In fact, numerous industry groups, including the Industrial Internet Consortium (IIC), Industry 4.0, the OpenFog Consortium, as well as multiple IIoT standards initiatives, such as the OPC Foundation, IoTivity and EdgeX Foundry, have emerged over the years.
"Most standardization efforts have focused on either rationalizing or harmonizing the profusion of data ingestion protocols in the various industrial IoT domains (factory automation, process automation, building automation), or creating a unified architecture for data management and communications," said David King, CEO of FogHorn Systems, based in Mountain View, Calif. The company provides edge intelligence software for industrial and commercial IoT applications.
IT and operations seen converging
While IIoT standards should result in faster, easier data transfers between devices and machines, the question is whether businesses are ready to jump in and coalesce around them.
"In the industrial space, what we're seeing is this convergence of information technology and operational technology," said Debbie Krupitzer, industrial IoT practice lead for North America at Capgemini. "And what's happening is the plants are adopting a lot of the IT best practices in terms of standards and platforms to get more consistent and [operate more cost-effectively]."
As for standards for industrial IoT platforms, it comes down to efficiencies and ways to aggregate data so executives can get more information out of their factories.
"We're seeing that the IoT standards are coming into the plant, which is a good thing," Krupitzer said. "The interesting part of it is that the things we do from an IT perspective don't always fit so well into the factory, so there's a little bit of a culture clash that's going on. What we're trying to do is get that balance of the standards, which is tying into platforms."
Because the standards are getting more elevated, there's also more visibility into the plants, and people are understanding why they should implement IoT. It's a great time for industrial IoT platforms, according to Krupitzer.
"This is why GE [with its Predix platform] has done a brilliant job of coming in at the right time, pitching that platform for that aggregation of all the factory data," she said. "What they're taking advantage of is the IT folks who understand platforms and systems, and now [they] want to apply that to the plant environment."
As the technology matures and more plants begin operating on digital platforms, standardization will become more prominent, Krupitzer said.
"It just takes a while, and then the standardization comes," she said. "You'll probably see standardization at the company level, and then you get the success stories, and then everybody wants to duplicate [that success]. This is especially true in manufacturing, where companies tend to have a show me attitude."
GE Predix, Siemens MindSphere, Bosch IoT Cloud and Honeywell Sentience are end-to-end cloud industrial IoT platforms that rely on Pivotal's open source stack, Cloud Foundry, for a platform-as-a-service layer, with back-end infrastructure provided by a combination of Amazon Web Services, Microsoft Azure and internally hosted data centers, King said.
Most commercial providers of cloud industrial IoT platforms have used open source big data infrastructure technology, such as Hadoop and Kafka, and they have used Spark, R and MATLAB to program data science analytics and machine learning models.
Out at the IoT edge, all of the large industrial players are using their own industrial controllers and gateways and working with x86 IoT gateway partners, such as Cisco, Dell and Hewlett Packard Enterprise, or ARM 32 gateway platforms from a wide variety of vendors, to deliver edge networking and security, according to King.
For edge analytics and machine learning, many of the leading IIoT companies are working with small, operations technology-centric software platforms that can run on different cloud services, or they're trying to run cloud-developed software on large servers or data centers.
A stack full of standards for manufacturing
While there are highly specialized sets of technologies and standards for various aspects of manufacturing, such as process manufacturing, there are no generic standards for the manufacturing sector overall, said Frank Gillett, vice president and principal analyst at Forrester Research, based in Cambridge, Mass.
"You have to get more specific about what aspect of manufacturing you're involved in," he said. That could mean specifying different parts of a company's supply chain.
One set of standards might center around track-and-trace technologies, such as radio frequency identification and near-field communication, Gillett said. And there are different standards for indoor locating and positioning, and different standards yet again for programmable logic controllers, which are common in factory automation.
The closest thing to an overarching IIoT standard is coming from the IIC, said Bill McBeath, chief research officer at ChainLink Research, which is based in Newton, Mass. The IIC is working toward developing interoperable IIoT architectures.
"They've put an awful lot of work into internet of things architecture, and also implementation guidelines," he said. "Those aren't exactly standards, but it's a common language and framework for thinking about the internet of things."
Additionally, three industrial groups, the OPC Foundation, OMAC and PLCopen, have started to work together to enable greater interoperability between the standards they've developed, with the goal of tailoring those efforts with the IIC's planned architecture.
"There are a whole stack of standards that are going to be necessary at some point, from the lowest level of connectivity to the highest level of semantics in various vertical markets. Like, how do you control manufacturing devices? And how do you control agricultural devices?" said Richard Soley, executive director of the IIC, which is based in Needham, Mass.
Standards at the bottom are starting to get adopted, but at the top, there's nothing, and it is going to take a while, said Soley, who is also chairman and CEO of the Object Management Group, an international, nonprofit computer standards consortium that focuses on semantic integration standards in a number of vertical markets, including manufacturing.
"The problem has been that there are already middleware standards just for connecting devices at the lowest level, but they're different in every vertical market, and IIoT is certainly not just manufacturing," Soley said. "The whole point is that IIoT is about connecting across markets, [and] using data in ways that you've never thought of before."
Although the IIC isn't a standards organization, it works with large users and vendors of IoT to IoT-enable industrial systems and learn requirements for new standards that the organization shares with open standards organizations, among other things.
Learn how IoT advances digital manufacturing
Read what process manufacturers can gain from IIoT
Get the handbook on supply chain applications of IoT