As enterprises rush to deploy AI across their factories and logistics sites, the less glamorous work of radio design has never mattered more. US network design firm iBwave argues that escalating AI workloads depend first on getting the private 5G foundations right – because there’s no intelligence without signal integrity, especially in complex and critical Industry 4.0 environments.
In sum – what to know:
RF before AI – iBwave says reliable AI performance needs reliable RF design first – because poor signal integrity undermines everything from robotics to computer vision.
Private 5G usage – private 5G usage has widened and deepened – across new sectors and new technologies; the latter trend sees increasing AI workloads on private networks.
AI as assistant – intelligence still needs engineers; iBwave is integrating AI into its design tools for simulation and analysis, while maintaining accuracy and human oversight.
Note: this is a feature-excert from a new RCR Wireless editorial report about private 5G and generative AI in Industry 4.0; the report is available here, and an attendant webinar session, also featuring iBwave, is available here.
There is one aspect of this conversation – about AI and 5G in Industry 4.0 – that tends to be forgotten, perhaps: that if you load your sparkling 5G network with spiralling AI workloads, to unleash untold automation in critical Industry 4.0 systems, then your network better be up to the job. “Because without quality RF design, none of this really works,” says Kelly Burroughs, director of strategy and market development at iBwave Solutions. “Every time you layer in new AI, whether for robotics or computer vision or generative AI, it puts more pressure on the network. And there are a few realities. One is that quality RF equals quality AI – which means maximum ROI for every application.”
Her company knows; if you’ve got five bars on your 5G device, deep in a hanger or deep in a mine, stuck between concrete walls or steel joists, then the chance is – probably, maybe – that US-based network design company iBwave helped out. “We may not be on the front line, but our software is used behind the scenes to design many of the private networks out there today,” says Burroughs. The company has been around for 20 years, offering site survey software for the design of all sorts of networks, from IoT sensor comms and enterprise Wi-Fi to public 5G and public safety systems. It has 4,000 customers, reckons Burroughs – a “good mix” of carriers, integrators, enterprises.
She says: “Our goal is to make the entire lifecycle of the network, including private networks, more efficient – from survey design to collaboration and management. There’s so much potential with AI, but if you don’t have a solid network foundation then it won’t work as well as it could. And as enterprises implement their private 5G and adopt generative AI, they also need more visibility and control tools at their fingertips – to manage their operations and work better with their providers.” Indeed, 20 years of design knowhow, and iBwave knows how network architecture has morphed and mutated in different industries with different technologies.
Burroughs comments: “Our viewpoint comes from our customers, who are designing and deploying these networks on the front line.” She splits the private 5G market horizontally and vertically, as it has widened across busy industrial sectors (utilities and public safety networks, to ports, factories, warehouses, and mines), and deepened across new technological advances (from simpler IoT sensor automation, to AI robotics and camera vision, to more futuristic generative and agentic AI). “Computer vision and robotics are the main drivers for private 5G, today,” she says. “But there’s some thought, as well, about how generative AI copilots can be used in these environments.”
She adds: “Enterprises are really just trying to set their networks for success with those main AI pieces, today – to be able to scale into generative AI in the future.” The framing is useful: private 5G is more than just hype, and firms with physical assets to monitor and optimise – vehicles in warehouses, robos on production lines, even baggage belts at airports – are the ones driving investments in 5G and AI. And she returns to this fundamental truth – that if the RF (radio frequency) design in the network foundations is not strong, then the whole AI house will come down. “That is the reality,” she says again. Optimise your network now, so you can build your new-fangled AI later – is the message.
Oh, and choose iBwave – is the implication. The firm is following its own AI path, too: using AI in its own tools – as it can; sensibly, a little cautiously – to help enterprises use AI in their factories and plants. Burroughs explains the discipline: “We are known for our accuracy, simulating the performance of the network. We don’t want to trade off that accuracy. But we are actively looking at ways to leverage AI to help users to accelerate these simulations – how to automate that piece, as well as to help with things like scenario planning. Because enterprises have a lot of tech options for different use cases, and a [quicker diagnosis to show one solution versus another holds value].”
She goes on: “The other thing is how we open access to all the data that sits underneath our platform – so customers can use their data [more productively] with their own AI applications in whatever ways they want. We want to put AI into the tool, but also enable our customers to use the data in other ways.” These points are interesting, and clearly relevant: the emphasis on simulation accuracy, as a ‘non-negotiable’ in high-risk industrial environments; and the recognition that data generated during network design has further value, for both the customer and the provider itself. It is a clear sighted view of the potential with AI, and of the risks, too.
Burroughs acknowledges its limits; iBwave is not about to hand charge of a private 5G project to a generative AI assistant, for example. “No,” responds Burroughs to the hypothesis. In plain terms: AI can speed things up, but it can’t yet replace domain expertise, she says – especially in industrial venues, filled with metal racks, moving machinery, reflective surfaces. “These are harsh environments. AI can help, even just with the KPIs to understand uplink or latency requirements. But we wouldn’t just let the AI run, and that’s it – go deploy. You need a detailed model of the environment. You need total accuracy. Maybe that might happen in the future, but not right now.”
In the end, the message is clear: start with quality design, survey, and modelling tools, and get the foundation right; Scale for AI later – and for generative AI when it actually works, and can be trusted. For vendors and integrators, generally, the AI opportunity is to combine domain expertise (in industrial networks, industrial systems) with advanced toolchains that automate where possible but don’t sacrifice accuracy or safety. The AI help is coming – but it won’t replace the engineer anytime soon.
As the industry transitions to true Industry 4.0 models, Burroughs and co are helping hidden networks to power visible outcomes. “We want to shift design tools away from being user-centric… to being more of a recommendation and analysis tool, and kind of a co-pilot to the design,” she says. If private networks are to handle AI co-pilots – and coordinated AI domain agents, at some point – then they need to be properly optimised, guided here and there by AI, but ultimately directed by human-made software tools. That’s the iBwave proposition, the message goes.
