I visited the opening of an industrial internet of things lab at National Instruments this week to see what is happening in the industrial side of the connected world. First off, I make a distinction between smart home, enterprise IoT and industrial IoT. I will likely define them in a later post, but for now, know that this event was focused squarely on industrial IoT.
Perhaps the most compelling demo I saw was associated with a standard that has been added to Ethernet called Time Sensitive Networking. The goal of the standard is to make Ethernet work well enough for the needs of industrial automation so it can replace the dozens of proprietary standards out there such as HART, CAN bus and more.
In dedicated networks, such as those for smoke detection systems or managing factory automation, data moves along an uncontested path to its destination. When we’re talking about information such as setting off a fire alarm or detecting a problem in a multi-million-dollar piece of equipment, delays can be costly.
The internet (IP) is different. It’s mostly best effort and information can get waylaid. That’s a function of its design.
In today’s factories, companies want to link their dedicated operations networks with their best-effort IP networks. That’s where timing can get screwy. You can’t transfer video (such as computer vision data from robots) and equipment health data using Ethernet without causing delays in your important equipment data.
This is why companies including NI, Cisco, Schneider Electric, Bosch and others are getting behind an IEEE standard called Time Sensitive Networking (TSN). The goal is to combine these two networks while still ensuring the important pieces of information from physical systems get to their destination without delay.
This is now part of the IEEE Ethernet standard so it is available in Ethernet chips hitting the market (and some that are already on the market). I saw it demonstrated at the NI IIoT lab and was impressed at how well this worked, but caveats abound.
First, this standard works by looking for the Mac address of specific devices and assigning a data path between them. So you have to know where you are sending your time-sensitive data. Second, it’s only ready for a subnet on a specific IP address, which means it’s for private networks today.
But if you’re crazy excited about this idea, then the Internet Engineering Task Force has a working group dedicated to trying to apply these concepts to the Internet at large. For more on that, check out its sections on deterministic networking.
There are clear benefits for this sort of blending in industrial networks, and other demonstrations in the lab showed off how this technology could help with things such as asset tracking and managing the unreliable flow of renewable energy back onto the electric grid.
The question is if TNS can help Ethernet make inroads into the factory floor despite its flaws. Jim Theodoras, who is vice president of global business development at ADVA Optical Networking, says yes.
“It seems like it can’t win but then they add so many extensions to make it fit, and when you look at the sheer cost differential, it’s got to go Ethernet,” Theodoras says. “They’ll have to do a lot to make sure everything works together, but it would be worth the pain just to get the economies of scale that Ethernet has.”
That’s why efforts like NI’s are so important. They let vendors guarantee that their versions of the industrial-IoT-strength Ethernet are interoperable.
“Nothing can stand up to Ethernet,” Theodoras says.