As networks evolve to 5G…Advanced and 6G, artificial intelligence is becoming a core component of RAN. But how can we test algorithms capable of learning and adapting to their environment without waiting for deployment in the field? This is the question that Rohde & Schwarz and NVIDIA are answering with their new test bench presented at MWC Barcelona 2026.
When the laboratory becomes a twin of the real world
Thanks to the integration of the NVIDIA Sionna Research Kit and differentiable ray tracing, the test bench recreates radio environments specific to a given site. Channels are no longer abstract: they reflect realistic propagation conditions, including site geometry, mobility and dynamic variations of the channel.
The laboratory thus transforms into a digital twin of the real network, allowing complex scenarios to be explored without logistical constraints.
AI at the heart of the radio loop
In this configuration, AI is not limited to a theoretical role. A link adaptation algorithm enriched by machine learning adjusts radio parameters in real time to optimize the spectral efficiency and robustness of the connection. It learns not only the channel characteristics, but also the behavior of user equipment, demonstrating the potential of AI‒RAN when tested in an end-to-end environment.
A collaboration that accelerates innovation

By combining RF instruments from Rohde & Schwarz with the computing power and software tools of NVIDIA, this platform creates a bridge between advanced simulation and experimental validation. It allows engineers to test earlier, faster and more realistic next-generation RAN architectures.
If you work on the mobile networks of tomorrow, this demonstration shows that AI‒RAN testing is entering a new phase of maturity. Digital twins and ray tracing pave the way for more precise validations, reducing risks in real-world deployments and accelerating the arrival of 6G.
Rohde & Schwarz | NVIDIA
Â




/2026/05/04/69f8366cb816b344604698.png)