Intelligent Networks for Time Critical Manufacturing

Excellent communication networks will be critical as manufacturing adopts 'Industry 4.0'. This is the fourth Industrial Revolution, driven by digitising the manufacturing process. Machinery function and performance, material stock levels, production inputs and outputs can all be monitored and controlled digitally, making use of technologies such as virtual reality, advanced analytics, advanced engineering and cloud technology.

This allows manufacturers to become more efficient and maximise benefit of systems like just-in-time manufacturing. However, these systems rely on digital networks for rapid communication and transfer of data. Whilst there are many possible ways for manufacturing equipment to communicate, (e.g.

WiFi), 5G is emerging as the best suited for data-heavy sectors such as manufacturing. It has faster speeds, higher capacity and lower latency, but even 5G suffers from delayed data transfer when the network demands are not prioritised correctly. This leads to costly delays in time-critical manufacturing functions.

TUBR are a UK SME, who are pioneering applications of _physics-based machine learning._ This subset of AI allows systems to use sparse and intermittent data to make valuable predictions. Usage patterns of private, industrial 5G networks present an excellent example of data sets with lots of points at some time and none at others.

Our aim in this project is to allow digital manufacturers to make better use of 5G networks by allowing flexibility around network rules called Quality of Service/Class of Service identifiers (QCI). Applying TUBR's machine learning will enable the system to modify QCI rules as 5G demand changes. TUBR will work with the Advanced Manufacturing Research Centre (AMRC) in the UK, the University of Duisburg Essen (Germany) and a German private 5G network company, Campus Genius, to apply physics-based machine learning to improve 5G network function through intelligent network control.