CD Laboratory for Digital Twin assisted AI for sustainable Radio Access Networks

This CD Laboratory aims to create a foundation for the use of artificial intelligence (AI)-based learning and updating methods in the field of wireless networks in various scenarios, which should benefit aspects such as efficiency, sustainability and reliability: To this end, so-called "digital twins" (DT) are being developed, which represent very different environments such as trains, industrial sites and dynamic environments, including the associated radio access and the respective user population in interaction with it.

Mobile networks have been an integral part of everyday and professional life for some time now, and they are also increasingly being used in industrial environments. Accordingly, the demands placed on them can vary greatly depending on the user, situation and application: Rail travellers want to work, communicate by phone or internet or use digital entertainment services during their journey in a fully occupied train. The speed, design of the railway carriage (metal and window materials that make it difficult for the signal to propagate) and the environment (if it is travelling through rural areas with poor network coverage, for example) pose major challenges for high-performance, reliable mobile communications coverage. The digitalisation of industrial processes, on the other hand, requires extremely reliable networking of machines in real time. These requirements demand more and more information from the mobile network, which also acts as a sensor.

The first generations of mobile communications were barely able to take differentiated user requirements into account during operation. It was only with the 5G standard that positions were implemented in the radio network that allow dynamic resource optimisation and enable so-called "network slicing". The challenge now is to develop methods for implementing this dynamic optimisation, and this CD Laboratory aims to automate the implementation of this using a model-based agent. This agent must always include the current status of the radio interface as well as a prediction of future load profiles. Successful implementation is therefore based on digital representations of reality - and this is where the digital twins, which form the basis of data-driven AI management, come into play!

Such a DT is to be understood as a mapping of a physical process or object in a natural environment onto a virtual object and enables validation, simulation or representation of its current or future status. A setup using traditional, purely data-driven machine learning methods would be a possible solution, but would require an enormous amount of resources due to the need for training data. However, by developing the various aspects of the respective network, including the environment and users, in the form of interactive digital twins of railway networks, industrial sites and dynamic environments that constantly interact with each other, the CD Laboratory team can use machine learning in various scenarios to improve wireless connectivity while conserving resources.