Steve Johnson, Ruckus’s regional director, Northern Europe, sets the scene: “Contrary to commonly held opinion, the UK is still the eighth-largest industrial nation in the world. It still contributes over 10 per cent of [the UK’s] GVA (gross value added). The UK’s manufacturing industry spends more on IT infrastructure than healthcare, education and retail, and it’s only just behind a tech-heavy investment industry like the finance industry.”
Manufacturing has also in many ways come full circle, returning to the time before Henry Ford pioneered mass production, when customisation to precise requirements was routine rather than a rare exception. The obvious example is the car industry, where the dizzying amount of choice is prompting manufacturers to invest in augmented reality (AR) apps to help their customers make sense of it all. According to Erik Josefsson, head of advanced industries at Ericsson, it’s also happening in the telecoms industry, with Ericsson’s own products being customised for the purchaser in a modular way. He explains that this trend and the demands it places on the factory floor in terms of flexibility is driving a shift towards mobility.
Similarly, Theresa Bui, director of IoT strategy at Cisco, notes: “Flex-scheduling for most factory owners now is everything. [To do it properly] you have to make sure that testing and calibration when you turn on your machine doesn’t take three or four hours when you expected it to take one.”
This need for real-time visibility is driving demand for sensors and the flow of information in staggering quantities – “a really well-connected digital factory will generate a thousand terabytes of data a day,” says Bui.
Josefsson adds that this trend and the need for reliable and stable wireless networks is generating interest in cellular. He also cites the issues that some manufacturers have encountered when using Wi-Fi at high densities, in terms of interference and their desire for predictable networks. “A cellular network is much more predictable and you can do preventative actions when it’s about to shut down.”
He explains that even current LTE can achieve latencies as low as 30 milliseconds and that while manufacturers must trust wireless technology in order to be able to go completely cableless, they are starting to look at 5G.
Given that manufacturing is a very risk-averse industry – “if they have to shut a factory down they know that they can start to apply for a new job very soon” – Josefsson says manufacturers’ use of cellular technology on the factory floor will begin with an on-premise network, and Ericsson is proposing this in its Smart Wireless Manufacturing offering. “So that even if something happens [to the service provider’s main network], the factory is going to continue to run on that local network. Then as the trust factor increases, we introduce network slicing and can guarantee the quality of service to a more central network. We start today with LTE and it’s going to be a path towards 5G. The use-cases that can be enabled initially [include] asset condition monitoring, and then [it will shift to] more positioning capability and AR/VR.”
Josefsson indicates that while AR/VR will start by providing fairly basic guidance to factory workers, “soon with the help of 5G, it’s going to become more of a human-machine interface where you move around with a tablet and you see the performance of the machine in real time and what’s going on inside with their tools – that requires extreme low latency and high bandwidth”.
He adds that the preferred model when it comes to accessing spectrum is for the manufacturer to use a mobile network operator/service provider.
Of course, many manufacturing facilities are already using cellular networks for more prosaic purposes, such as allowing their staff to make calls – and as Klaus Allion, managing director at ANT Telecom, points out, this isn’t always possible due to coverage issues, in which case, two-way radio may be the best option.
He notes while working with MNOs to improve the coverage on a large scale is an option, customers who have explored that route comment on the high costs and being tied in to the provider; ie, “in two years’ time when your contract runs out, the provider knows they’re the only ones you can get site-wide coverage from and can use that in their negotiations with you”.
At a smaller scale, “there are signal boosters out there”, but Allion notes there seems to be confusion around the legislation of using these devices. He adds that ANT Telecom can also provide customers with a private LTE network, “but that hasn’t found much uptake yet”, citing one customer who preferred to “stick with what they knew, which was DECT”.
The advantage of being able to use a smartphone over a DECT phone or a radio is highly attractive for some. “If I had to guess [why they don’t opt for private LTE networks], if it’s not cost then it would probably be not knowing exactly how the technology works and how customers can use the same smartphone on-site and off-site – which is possible with this solution, but not with other solutions like DECT or radio.”
Growing pains
Josefsson adds that some of the industry partners that Ericsson is speaking with have jumped onto technologies very fast, installing separate systems for Bluetooth, Wi-Fi and LoRa “at the same time as they’re leveraging maybe some incremental system. Some of them complain that it feels like the roof is falling down because there are so many different types of access technologies and they all serve so many different purposes.”
Similarly, Cisco’s Bui says IoT has resulted in a huge amount of complexity, which can make extracting data from devices a challenge. “[Say you have] 11 [types of sensor] on your factory floor from different sensor manufacturers. That means that all of them probably have different data models, data schemas, so the extraction of that data is costly and resource-intensive.”
There is also the difficult task of integrating new sensors with pre-existing assets, which may already have some degree of custom implementation. “Even though this is one of the most advanced verticals in the world of connected devices, most of our customers still rely on custom-coding projects; to really pull the efficiencies of data from their equipment, it still requires some downtime on their equipment, which they can ill-afford.”
Bui also sees many factories where there are silos of automation – “they’re not holistically connected end-to-end, all the way from the beginning of production to the finished product”.
It’s worth noting there can be limits to automation, as seen most recently with Elon Musk’s decision to reduce the amount of “excessive automation” on Tesla’s Model 3 production line, with one analyst at Bernstein, Max Warburton, who previously worked at the International Motor Vehicle Program, saying “automation in final [car] assembly doesn’t work”.
Bui adds that another issue is the operational side of the business often buys sensors or machinery to improve yields or reduce energy costs but without thinking about network latency, which can cause headaches for IT departments and can disrupt the real-time flow of information. Finally, she has seen that most companies still tend to rely on their operator’s experience and gut instinct rather than use predictive analytics, “but they’d like to move away from that”.
OEMs: leading the charge
Josefsson says many original equipment manufacturers (OEMs) look to 5G to reduce the intelligence in their devices by moving it to the cloud, which will reduce their cost, change the business model from capex to opex, and pave the way to a large ecosystem of industry apps running on them and the rapid introduction of new services. He adds this scenario covers some of the most aggressive use-cases for 5G in terms of high bandwidth and low latency.
Bui says one headache for manufacturers created by this shift to the cloud is the potentially limited control it gives them over where their data is going and the risk they may be inadvertently sharing proprietary information. Manufacturers need to send data back to the OEMs so the latter can perform remote monitoring and diagnostics on their equipment, but in doing so the OEM might be able to determine the extent to which production lines are being utilised or how many layers of paint are being applied to a car – “which is incredibly proprietary information”.
Most of the manufacturers Bui speaks with want the ability to split their data streams into categories and be able to choose which stay on premise, which go to the OEMs and which go up to the cloud for historical analysis. This can be done with platforms like Cisco Kinetic. Bui adds that where service level agreements exist between manufacturers and their OEM suppliers, OEMs are starting to put in cellular connections as a back-up so that in the event there is a network failure on the factory floor, they can still get the data they need to meet the SLAs.
Bui says: “It’s the device manufacturers that are really leading [on new technology in the manufacturing sector], meaning the factory owner gets what they get.” She gives the example of the transition from cabled to cableless devices, which OEMs are pushing for because it makes it “easier for them to install, maintain and monitor their products”. While Bui says there isn’t going to be a single connectivity choice for manufacturers, as far as low-power wide-area (LPWA) technologies are concerned, NB-IoT has really taken off in Asia.
Wi-Fi
Wi-Fi also plays an important role in factories. Just as LTE isn’t standing still – with the evolution of 5G – Johnson says Ruckus believes Wi-Fi will increasingly be about enabling the Internet of Things, through acting as the backhaul for a range of LPWA technologies such as Bluetooth Low Energy and Zigbee.
“Where you are deploying a Bluetooth-based IoT sensor in your manufacturing plant, rather than needing to connect back to the network and manage a completely separate and independent network to enable that data to get to the application on the server, that Bluetooth sensor can attach to one of our access points where we have our IoT suite deployed.”
He adds that while some manufacturers will wish to manage their Wi-Fi networks themselves, he expects Wi-Fi as a managed service/private cloud to grow faster. He says: “Every vertical has its own demand, challenges and opportunities, and very often those are shared
across a set of customers in the same market, and manufacturing will be exactly the same.”
Duty of care
Manufacturers also rely on wireless comms for push-to-talk (PTT), lone-worker applications, critical alerts and man-down applications. ANT Telecom’s Allion notes that when it comes to voice, there are typically two main modes of communication in a manufacturing context, “an almost command and control quick instruction type” – which is best served by radio – and one-to-one, which is typically handled by smartphones/DECT phones or landlines. In this context, he explains, there is little need for PTT to work across different platforms. However, where integration is far more important is “when it comes to lone-worker and critical alerts”.
Allion says companies want to avoid having different lone-worker scenarios for users with different devices, and the ideal is to have one control room where all personal staff alerts can be seen and managed, “and the same applies to critical alerts”. If a critical alert from an industrial process comes through, the engineer responsible needs to know regardless of whether they are on site or not, so they must be sent to multiple device types (such as two-way radios for on-site personnel and smartphones for those off-site).
He feels that automatic man-down detection still needs a bit of fine-tuning in an industrial context – because it’s possible to trigger it while crawling under a machine, and sometimes people have to work without moving around. This creates the danger that the number of false alerts/queries to the user could make them deactivate the feature and then suffer an accident. To reduce this issue (which can’t be completely eliminated), ANT discusses it with customers and tries to strike the right balance between detecting incidents and not frustrating their workers.
We have seen modern manufacturing methods and the need for flexibility on the factory floor are leading to a data-tsunami that creates problems of its own, and which while not insurmountable need to be addressed to prevent them from undermining the underlying business case. At the same time, as Josefsson points out, there is “a little bit of gap between the vision and the execution”. As that gap starts to retreat, it will be interesting to see how this shift will affect businesses and consumers alike.
Practising what you preach
Ericsson’s Josefsson says his company is using NB-IoT and AR in its factories. “For example, in Nanjing, we had quite a manual task walking around and looking at when tools needed to be replaced. Now, we have NB-IoT sensors that measure the torque, when it’s time to maintain them, [allowing us to be] proactive and fix screwdrivers when needed.” He adds that this project broke even in just six months and reduced the “manual, quite intensive, boring work by 50 per cent”. As for AR: “We introduced our own troubleshooting tool with AI in our [Tallinn] factory. Before you had to go through quite a long [list of] manual and testing procedures, now you just put in an AR interface which then identifies if there’s any faults compared with a perfect circuit board and directly guides you through that journey, which also improves efficiency by around 50 per cent.”
