5G and AI – an inseparable couple

With Industry 4.0, the amount of data on the factory floor has increased significantly. More and more companies are collecting and using this data to enable innovative services around production and manufacturing, such as predictive maintenance. The continuous tracking of operating data, such as mechanical and thermal stress, in combination with the automated checking of the quality of manufactured products, enables many manufacturing companies to optimise maintenance cycles and plan them more precisely.

But manufacturing and transport processes are also generating more data – including continuous video streams, for example from automated guided vehicles (AGVs) or robotic arms whose activities are monitored by cameras – whether as part of the work process or for occupational safety reasons. These application scenarios in such diverse sectors as transport and logistics or industrial products have one thing in common: the evaluation of the data, which results in the output of recommendations or autonomous control, falling into the broad field of artificial intelligence.

AI needs resources

In the “smart factory”, logistics centres, and the transport or retail industry, higher storage and computing capacities are therefore needed more so than in the IT environment of days gone by. Instead of equipping the individual machines, vehicles and robots with ever more powerful controllers (PLC) or industrial PCs, it may be more economical to outsource the AI functions. Edge servers provide the required computing and storage capacities more efficiently and cost-effectively at the edge of the shop floor, can be scaled flexibly and are easier to manage. On the machine side, space and material are saved. In the case of mobile equipment such as conveyor vehicles or robots, less mass to move also means a reduction in energy requirements.

At the same time, for more and more companies, edge servers are an alternative to distant public cloud solutions whose availability is not sufficiently guaranteed in combination with failure risks in the WAN area for manufacturing and production processes. In the meantime, numerous server providers have adapted to the requirements of corporate customers and make suitable offers within the framework of a MEC concept (multi-access edge computing) – often in cooperation with a network manufacturer. Because the MEC concept in industry stands and falls with reliable and powerful data communication between machine and server.

For mobile applications that require wireless data transmission, a 5G network creates the necessary conditions. High bandwidth, low latency and extreme reliability characterise the new standard. In addition, there are advantages for companies such as better physical propagation even in difficult environments where, for example, undesired radio reflections are caused by metal pipes or containers, or the unproblematic handling when changing the radio cell when autonomous vehicles are used on a larger company site. Wireless networking via 5G also lends itself to situations where production or logistics environments need to be adapted more frequently to different job scenarios. And last but not least, it can serve as a resilient backup that is just as powerful as a cable-based infrastructure.

Network integration is crucial

What the technology itself can achieve is one thing – but how the results look in practice is something completely different. In particular, it depends on the interaction of the different network components. For example, in order to be able to closely integrate an AI application on edge servers into manufacturing and production processes, companies first need a suitable system architecture on the server side. Based on this, the next step is to integrate it into the (wired) core network of the company as well as into the RAN infrastructure of the radio network. In this way, the entire infrastructure can be efficiently managed and its performance can be permanently guaranteed during operation.