Alexander Pil
10 February 2021

With a response time of a millisecond, a 5G network makes it possible to perform advanced control tasks, hosting computing power in the cloud. Very attractive for industrial automation, but it’s far from easy to get such a network up and running. Network operators still have little understanding of the requirements of the manufacturing industry. Fraunhofer IPT’s Niels König is working on a solution.

5G is hip and happening. From all sides, consumers and businesses are bombarded with messages advertising the latest generation of wireless technology. Is there a hefty dose of opportunism among network operators, and a solid technology push? “Here in Germany, companies can apply for part of the spectrum themselves and don’t necessarily have to collaborate with an operator,” says Niels König, department manager of production metrology at Fraunhofer IPT and coordinator of all 5G activities at the Aachen institute. “Those parties indeed have to push to get their offerings to the market. They then join forces with an appealing name like BMW or set up a 5G network in a soccer stadium. I find that very marketing-driven indeed.”

Traditionally, of course, network operators have been heavily focused on the consumer market. “That’s where they come from, where they’ve been making a living for decades. Their whole organization is set up to sell SIM cards. With 5G, they’ll attract few new customers in that sector; for consumer use, the new generation is a normal evolution compared to 4G. Of course, the speed will increase significantly, but all network operators can use that same argument.”

Fraunhofer IPT Niels Konig DSC_6439
“As an industrial user, to benefit from the advantages 5G offers, you definitely need a private network,” says Fraunhofer IPT’s Niels König.

König wants to clear up a major misunderstanding right away. The great promises of 5G – a throughput of 10 Gb/s and a latency of 1 ms – are only achievable on a private network. “You really won’t achieve that on your smartphone,” he argues. “With a new 5G subscription, you’ll probably have a solid performance, with data rates of hundreds of megabits per second. But once the number of users increases, you’ll have to share the bandwidth with more people and as the data volumes get bigger and bigger, the performance will go down.” Consumers may be able to live with that, but for industrial users, it’s unacceptable.

“A standard 5G network is also optimized for downlink, so consumers can reel their Netflix series in quickly,” König continues. “It’s much less good at uploading, which is exactly what the industry needs. Companies want to send their data to an edge device or the cloud. As an industrial user, to really benefit from the advantages that 5G offers, you definitely need a private network, in a configuration that suits your needs.”


Device lifecycle management for fleets of IoT devices

Microchip gives insight on device management, what exactly is it, how to implement it and how to roll over the device management during the roll out phase when the products are in the field. Read more. .

Clash of worlds

Such a private network sounds simpler than it is. Indeed, network operators have little understanding of the industry’s requirements. “After years of focusing on the consumer market, it’s also not fair to expect them to suddenly understand production now,” König believes. Fraunhofer IPT is one of the participants in the 5G-Acia consortium (Alliance for Connected Industries and Automation), which aims to roll out 5G across multiple industries. “The operators are also in there. It regularly turns out that the different parties don’t even use the same vocabulary. It’s a clash of two worlds.”

The industry is a completely new market for network operators. They know perfectly well how to serve consumers. However, if you want to provide services to industrial customers who have installed their own 5G networks on the factory floor, that’s something completely different. Each configuration is slightly different, which means that operators have to spend a lot more time on it,” says König.

Mercedes-Benz officially opened its state-of-the-art Factory 56 last September, which is going to produce the company’s S-Class. The factory is equipped with all sorts of high-tech gadgets, including its own 5G network set up in collaboration with telecom giant Telefónica and network supplier Ericsson. What happens if that network falters or even breaks down completely? If that means production stops, can Mercedes recover those costs from Telefónica and/or Ericsson? “Such a business model is very unconventional for network operators,” says König, “and requires a drastic change in their market approach.”

Chicken or egg?

There’s an additional problem: 5G is only one piece of the digitalization puzzle that the industry needs to put in place. After all, 5G is just the connectivity. But what do you do with the vast amount of data collected on the factory floor? Where will that data be processed? And how do you deal with cybersecurity? “All of that is a much bigger topic,” argues König. “How do you build a control loop between the machine in the factory and the data processing in the cloud, for example? The success of industrial 5G hinges on whether we can solve this in the industry itself.”

Deutsche Telecom now has a systems integration branch, T-Systems. That’s a step in the right direction, König believes. He also expects a lot from external companies that act as intermediaries. “There are already a few of those now, having smaller networks and providing customized services to industrial customers. Almost without exception, they come from the telecom world. None has a great track record in industrial networks.”

For the large-scale acceptance of 5G in industry, König also sees an important role for research centers such as his own Fraunhofer Institute. “It’s a chicken-and-egg issue. You need a market pull with companies asking for a private network, equipment and services. From the research side, we can strengthen that pull by pioneering. By showing that it works and what it delivers. And by identifying the right applications. The use cases we’re developing now should convince potential customers.”


In the past year, Fraunhofer IPT has installed a lot of 5G equipment at its research centers in Aachen. “We now have four 5G networks fully operational,” states König. “We need these different networks because we’re dealing with several research institutes and each research project requires slightly different network characteristics.”

König himself is most closely involved in research projects around machining. For example, he and his colleagues are working on a very demanding industrial application: the milling of integrated turbine blades. These so-called blisks (blade-integrated disks) are components of the compressors in the jet engines of client MTU Engines. Until recently, they could only be inspected after a forty-hour milling process, with the costs of failures due to defective products or rework amounting to 25 percent. König: “Blisks are made of special alloys and are very expensive, in the order of forty to sixty thousand euros for products in series. We estimate that by using 5G process control, the annual savings for one blisk factory could be as much as 27 million euros.” The 5G network and the first-time-right approach at Fraunhofer can soon be adopted by MTU. “The way we’ve set up our 5G network, I think, is also a good example for other industrial users.”

Fraunhofer 5G blisk machining
Fraunhofer IPT uses 5G process control in the milling of integrated turbine blades. Credit: Fraunhofer IPT

Other research projects revolve around mobile robotics and automated guided vehicles in production logistics. Accuracy is an important issue there because those systems are even less precise than static robots. “You need sensors to tell where such a mobile robot should pick up something. If you use it for inspection, you don’t want to do the image processing on the robot itself; that costs too much energy. It’s more convenient to send the images to an edge device or the cloud. Then you have a sensor in the loop. Also, if several robots work together, for example around the production of a part of an aircraft fuselage, you need to control the data flows. In all these kinds of applications, 5G can help you.”

Network operators are touting their 5G networks as being super fast (10 Gb/s) and with a latency of ultimately about 1 ms. König: “That first characteristic isn’t very interesting for industrial applications. You have to be very imaginative to think of an application where you want to send 10 Gb/s of data from the factory floor. This is just as well since the two characteristics clash. You have to choose high speed or low latency; it can’t be done at the same time. Because few industrial parties need super-high speeds, they can aim for optimal response times. And if you don’t need that low latency, you might as well use Wi-Fi.”

It’s precisely that low latency that makes 5G a game-changer in the industry. In a private 4G network, you can get a latency of about twenty milliseconds. Pretty good but cycle times in advanced automation tasks are well below that. 5G is the first communications standard fast enough to handle those kinds of processes.


The current state-of-the-art doesn’t yet reach that target of 1 ms. Fraunhofer IPT is now using equipment based on release 15 of 3GPP, the global alliance of standardization organizations responsible for 5G, among other things. The rules for release 16 were adopted only last September, including an ultra-reliable low latency of 1 ms. The first products based on that standard will appear on the market this year.

Fraunhofer 5G Smart Demonstrator
5G is the first communication standard fast enough to keep up with cycle times in advanced automation tasks. Credit: Fraunhofer IPT

“With our current systems, we achieve a latency of five to seven milliseconds,” says König. “For many applications, that’s enough. Consider that for safety, communication between an AGV and a command center or temperature control, you don’t need 1 ms.” König sees little added value in remotely controlling his milling machines – an application where latency is crucial. “It’s primarily process monitoring. To this end, we’ve developed wireless sensors at Fraunhofer that can record acceleration and acoustic emission during machining. This is interesting in, for example, a five-axis milling machine with a lot of movements. You don’t want all kinds of cables there. You can easily use these sensors in an existing system, as a retrofit. You don’t have to completely change your system and the advantage is immediately clear. Especially with critical, expensive parts, like MTU’s blisks, you want instant response. With their low-latency properties, our 5G sensors do well there.”

The next 3GPP release, 17, also still counts as 5G. From release 18 onwards, it’s 6G. “It’s striking to me that many people already talk about 6G, while 5G hasn’t even been properly rolled out yet,” König concludes smilingly. “I expect that in a year or two, we’ll have reached the point where 5G will be normal in the industry.”