‘A thousand repeats of the same conversation’ – four challenges to make private 5G scale

There are a whole bunch of headaches, of course, that come with scaling private 5G systems to new venues and new markets. Even if they have been proven at one site, and the use cases and business cases have been mastered, the challenge to easily replicate them in new industrial domains is invariably fraught with local chaos and stress. This was the subject of a panel discussion at Industrial 5G Forum (available on-demand) last month, chaired by Analysys Mason, and attended by Druid Software, Hub One, and Three Group Solutions – all variously deploying and testing private 4G/5G solutions in Industry 4.0. 

Three Group Solutions talked about what it had learned from a decade of private networking knowhow at Hutchison Ports; the two companies share a parent, in the form of industrial conglomerate CK Hutchison, and its telecoms business, with 11 mobile operating companies around the globe, has been supplying sundry 4G/5G solutions to its ports business, the third largest container port operator in the world. Graham Wilde, head of 5G business development at Three Group Solutions, kicked off the panel by racing through the standard obstacles to scale: standardised security and common applications. 

He said: “The more sites you have, the more security issues you have. There’s no one-size-fits-all answer to solve that security problem, and it takes a lot more thought… when you have multiple locations… You [also] have slightly different use cases in each place. Which sometimes drives different equipment [choices]… to connect to the network [which] drives different solutions.” Both of these points seem self-explanatory, and are linked invariably to issues just with the variety and variability of workloads at different industrial sites. But he also hit early upon a more prosaic infrastructure challenge, as experienced with Hutchison Ports over the years: backhaul. 

1 | BACKHAUL 

On backhaul, Wilde explained: “The number one issue until now has been the reliability of moving the traffic from one place to another place. It’s actually nothing to do with the core or the radios; it is about the physical fiber connections, leased lines, VPNs, whatever – which shift traffic from one port location to another… It’s not very sexy; it’s not something people talk about very much in private networks… but it is a real issue, and something that’s bitten us in the past.” His firm has been running a single private 4G/LTE network across two locations for Hutchinson Ports for 10 years; it is in the process of being upgraded to 5G, and expanded to three sites.

But the issue, said Wilde, has been to do with the setup of the industrial sites, across two sides of a river, and the vintage of the original LTE network solution, which was built before localised private cellular core networks could be easily distributed (“split”) across multiple sites. “You had to have the core in one place and if a site lost connection to the core, it would just stop working,” he said. It is different now, he observed. “Now you can split those core functions, and keep the private network running – even if you lose backhaul connectivity.” But the solution, engineered at the time, has been to run a leased-line connection, supplemented with a microwave link, across the river.  

He said: “The least reliable link in the chain is these connections from location A to location B, where the core capability sites… Even though we were buying physically diverse routing between A and B, the reliability of the links was still not very good, and the time to fix them was very poor… [And] because there are big container ships going down the river, the microwave link sometimes gets blocked by big ships. So we put another microwave link in, and… we are [now] looking at low-earth orbit (LEO) satellite as another way to back up connectivity between points A, B, and C.” Simple backhaul, the story went, can be a major problem, if it is not solved from the start. 

2 | SPECTRUM

The story is anecdotal, but also apocryphal. Planning matters most, and it came out in every response – notably when advising on how to navigate the cascading challenges that come with fragmented global spectrum regulation. “Different geographies, different regulations, [different] frequencies [are issues],” summed-up Gregoire de la Crouée, business line director for private 4G/5G at Hub One, which is owned by Paris Airports Group (Groupe ADP), operating 26 international airports, and is selling private networks as part of an Industry 4.0 integrator offensive to various industries. “On top of that you also have the issue of the bandwidth – the width of the spectrum – which is not the same [everywhere].”

De la Crouée explained: “It’s not the same to have just 20MHz [in one country] and 100MHz [in another] for different use cases… [And] the process [and time] to get those frequencies is different as well. So you have to cope with the local constraints… There is an impact… on the radio infrastructure, as well – which will be different in band N77 and band N38, for example. So you have to find the right equipment… [and also the right] devices… Devices have to be fully adapted, and the maturity of the device ecosystem is really different.” Different spectrum requires different radios and devices, often; but there is a classic chicken-and-egg challenge with devices, as always. 

3 | DEVICES

But David O’Byrne, senior business development manager at Druid Software, asked for calm, effectively, and suggested the ‘device’ ecosystem – which stretches in the Industry 4.0 space into connected industrial equipment and machinery – is going at an industrial pace, and also developing nicely. He said: “It’s important to remember that this goes at industry-speed. The whole market is still in its infancy. If you’re from a telco background, it’s important to learn to speak like ports and warehouses, and manufacturers. Previously, the telco industry just talked telco, where a new 3GPP release comes out every year or two, and devices come out very quickly.” 

He went on: “Private 4G/5G is an enterprise and industrial play. It’s not a telco play. So we have to learn their language, and understand that device cycles and availability are much slower. The first chips [for] a lot of private 5G devices were only made available in 2023… These devices will be current for a decade. That is the mentality shift… Device availability is improving, as well – on a monthly basis… Modules became available this year, [and] are now being tested.. It’s going to be a 30-month cycle… The telco world needs to understand that this is not slow. It is the appropriate speed for people who make cars and ships, and move containers around the world.” 

There was a call from de la Crouée for device makers to enable voice over LTE (VoLTE) in private 4G/5G bands. And O’Byrne explained how the ecosystem must pull together. “It’s not just the Apples and Samsungs; it’s router manufacturers, forklift manufacturers, crane manufacturers, all sorts of robotics companies,” he said. “There is an entire ecosystem – [companies] that make handheld devices [just] for surgeons. That’s all they do; one hundred percent of their business. The challenge… is to talk to those people and say, okay, how can we make yours a cellular device? It’s a thousand repeats of the same conversation, but it’s the only way it gets done.”

Again, his point was about telcos talking-the-talk’in Industry 4.0.

4 | MANAGEMENT

They also need to talk-the-talk in local markets, with local enterprises and local regulators; this is invariably done via local integrators and specialists. De la Crouée explained: “Lots of our customers want a single point of contact for all their geographies…. So you have to find local partners… to adapt to local constraints. [Which is reflected in] the architecture strategy as well – [which should] take account of data sovereignty…. You can imagine a core network located at a single point of contact, but, then, how do you get the data outside of [a country or region]? Do you keep it all for yourself, or have it managed elsewhere? That’s a real security issue.”

As an aside, before getting back to the point, Wilde suggested enterprises and integrators should simplify their systems by reducing the variety of different vendors and devices in them. Don’t mix, unless you have to – was the message. He said: “Stick to one or few vendors across all your systems… We keep the ecosystem quite restricted so it makes the management easier… A limited set of devices [is] also helpful – because devices don’t come from the same company as the core and the radios, and you’ve got to spend time integrating them into a single management system. Wherever we can, we keep it simple – by reducing the number of vendors in the system.”

He related this to an active management scenario at a port, where Three Group Solutions has been called upon to fix automated guided vehicles (AGVs) – and close integration with the device (AGV) vendor has afforded it a fuller view of the setup, and a clearer way problem-solve. “When one of those [AGVs] stops working, [the customer doesn’t] know whether it’s the vehicle, the management system, or the network. But they call us. It’s not always enough just to have a view of the network. Having an application level view is also useful. So we do work quite closely with the vehicle suppliers so we can pull [application level] elements in,” he said.

Last word, here, to O’Byrne, who flipped Wilde’s message about rationalising vendors to urge enterprises and integrators to relax about engaging contemporary private multi-RAN networks. “We have made flexibility our strength. Our [system is] built [to] provide APIs to any orchestration platform. We leave that decision to the customer or integrator. Our [proposition is] a comprehensive API that allows you to take any event on the network and plug it into your orchestration system. Maybe it’s because we are a relatively small company, but we have gone completely the other way and made flexibility and the ability to work with everybody, our hallmark in that area.”