Connectivity is Key for Success at the Industrial Edge

As manufacturers delve into digital transformation efforts that aim to move computing and machine learning technologies from cloud environments closer to the machines processing critical data, lots of attention is paid to the infrastructure such as controllers, edge servers and Internet of Things (IoT) devices. However, ensuring that data from those devices can properly reach the computing resources requires focusing on the right networking technologies, whether wired or wireless.

"Connectivity is the nerve system of the factory," said Sander Rotmensen, Director of Industrial Wireless Communications at Siemens. "If you look at all of the assets on their own, they’re all great things, but if you have an edge device that is unable to gather data from the field or not able to communicate with the cloud system, you have nothing. With next-generation technologies, we really can tap into a new level of low latency and ultra-reliable communication that gives innovation in digital automation an additional boost."

Related Webinar: Convergence of Connectivity and Compute Unleashes Intelligence at the Industrial Edge

Watch for Free

Rotmensen was one of the panelists in the recent A3 webinar, “Convergence of Connectivity and Compute Unleashes Intelligence at the Industrial Edge,” which discussed how connectivity and artificial intelligence would derive exponential value from the data created by billions of connected things. The webinar was part of the Intelligent Edge for Industrial Applications series, sponsored by Intel. Rotmensen was joined by Intel’s Mukund Shenoy, Director of Industrial Connectivity Strategy.

Shenoy explained that traditional infrastructures within the factory are moving to an edge approach that involves taking data from field equipment, machinery, and other locations and processing them locally on a private cloud instead of moving to a centralized data cloud. "All of these systems are unlocking data that have been trapped in the past systems," said Shenoy. "Now that data has to move to a place inside the four walls of the factory, through a local automation cloud where you can provide more compute power to crunch the data, apply machine learning and AI, which can then be scaled up and down. But the pivotal thing happening is the connectivity pillar becomes more important because the data has to move fast."

New technologies such as AI, ML and robotics require low latencies and high bandwidth, which can be achieved through new connectivity choices such as 5G wireless and Wi-Fi 6, to name a few. During the webinar, Shenoy and Rotmensen discussed the benefits and advantages of both approaches (Wi-Fi is easier to set up and deploy, while 5G is better suited to connect multiple applications in one network simultaneously.

Both panelists agreed that companies should explore private 5G networks for their manufacturing facilities rather than relying on public 5G networks. "Would you really trust your factory to be run on a public network or connected to a public network?" asked Rotmensen. "If I walk in my door, my signal drops from 5G to 4G, and when I’m downstairs in my basement, it’s completely gone. I think having a private network set up exactly where you need it, getting the exact requirements you need from the network, is something that is really helpful. The same situation exists in the factory. You can bring the wireless signal everywhere, in every nook and corner of your factory, even when there’s a lot of metal around it. This helps you create a reliable network to where you want to have those low latencies."

Matching the network architecture to the compute needs is also important. The network should be planned to ensure you know who is on your network. The network should also have enough radios to cover the factory to avoid situations in which a cluster of end devices is connecting to one radio, while the other radios are doing nothing, Rotmensen added.

When asked whether manufacturers should trust wireless networks from a security standpoint, both panelists agreed that standards organizations within the technologies, such as the 3GPP (3rd Generation Partnership Project) for 5G, are working hard at establishing standards that manufacturers can trust. "The ecosystem is rallying around delivering products for industrial uses, with a solid contribution from standards bodies," said Shenoy. "The benefit of 5G with its bandwidth and latency is not so much driven by consumer applications. There is only so much you need on your phone, right? This technology targets industrial uses. For that reason, we are both part of a consortium of various industries from all over the world to jointly provide inputs and push the development of the 5G standards forward, the 5G-ACIA (Alliance for Connected Industries and Automation). The OT, IT and communication technology industries are collaborating to ensure that we have robust industrial strength devices that can stand behind all these systems with performance guarantees."

While the panel discussed much about wireless connections, they also reminded the audience that they shouldn’t simply replace wired connections. While many new applications entering a factory, such as autonomous mobile robots or automated guided vehicles, rely on wireless connections, many stationary pieces of equipment can still be connected via wired technologies. While many new factory installations (aka "greenfield") would likely adopt a wireless connectivity approach, existing factories that already have wired connections ("brownfield") could still benefit from an edge strategy for devices that might not yet be connected.

Regardless of the connectivity option chosen, the panel agreed that extracting data from the machinery and IoT devices deployed in a factory can help manufacturers become more efficient in their operations. As a final piece of advice, the panel mentioned being open-minded to the possibilities of the intelligent edge, aligning the operational strategy with the business strategy, and taking baby steps in the process.

"Don’t look at what is not possible," said Rotmensen. "Look at what is possible and reach out to the experts in the field." "There is a long learning curve here, and I think companies should start experimenting," Shenoy added. "Initial fast failures are something you can learn from, but also reach out to people who have already gone through this journey to help steer you."

To watch the full webinar, click here to view it on demand.

The next topic in the Intelligent Edge webinar series, “The Transition to Software-Defined Control Systems,” is scheduled for Thursday, Oct. 18, from 1 p.m. to 2 p.m. ET. You can register for free here.