NVIDIA donates 10,000 GPU hours on A100 to our professorship for RF simulation of satellites

We are proud to announce that NVIDIA has selected our project “Radio Frequency Physical Layer Emulation for Satellite Megaconstellations” for including in the Acadmic Grant Project.

With respect to this, NVIDIA donates 10,000 hours of managed A100 GPUs in the cloud for our research.

We are working on a big data challenge that arises when one tries to build a digital twin of next generation non-terrestrial networks. Not only do such netwowrks employ thousands of largely autonomous satellites serving complex base stations, the ground segment is as well mobile and complex with different numbers of users in different regions. Classical approaches to this simulation often compute link budget statistics and work from there, we want to perform a full time-discrete simulation in order to be able to correctly estimate channel parameters.

When successful, we might as well want to add wheather data (e.g., clouds) to the model in order to better understand our future digital communications systems.

It is worth noting that this approach is not only relevant to communication satellites, but as well applies to any constellation of Earth Observation satellites in a similar way. Only with a full digital twin, sensor parameters and communication capacities can be understood reasonably.

We are currently collaborating with Bundeswehr University and a major industrial partner in this area in order to perform real customer-driven analytics.