Nvidia ramps up 6G RAN ambitions with new research platform

During its annual GTC AI event, Nvidia took the wraps off its new 6G Research Cloud Platform, a three-pronged approach to putting itself right in the thick of the action.

Prong number one is a digital twin that can accurately simulate 6G systems, from a single mobile tower to a city-scale deployment. Called the Nvidia Aerial Omniverse Digital Twin for 6G, researchers can use it to build base station algorithms based on site-specific data, which can train AI to improve transmission efficiency in real time.

The second prong is Nvidia's Aerial CUDA-Accelerated RAN, a software-defined, full RAN stack that researchers can play around with – customising, programming, and testing 6G networks – also in real time.

Last but not least is prong number three: the Nvidia Sionna Neural Radio Framework. This enables developers to use Nvidia GPUs to generate and capture training data – including data generated by Nvidia's Sionna 5G and 6G physical layer simulation tool – which can be fed into deep learning models built and developed on the PyTorch and TensorFlow deep learning frameworks.

"The massive increase in connected devices and host of new applications in 6G will require a vast leap in wireless spectral efficiency in radio communications," said Ronnie Vasishta, SVP of telecom at NVIDIA. "Key to achieving this will be the use of AI, a software-defined, full-RAN reference stack and next-generation digital twin technology."

This latest raft of new solutions, packaged together as the 6G Research Cloud Platform, could give Nvidia a more fundamental role in the development of next-gen mobile tech, rather than a supporting role. It also gives Nvidia an opportunity to showcase its capabilities to the telecoms industry.

On that note, Nvidia reeled off an impressive list of first adopters and ecosystem partners.

Nokia, Softbank, Samsung and Fujitsu are perhaps the highest-profile telco names on this list. Test and measurement firms Viavi and Keysight are involved too, as is kit maker Rohde & Schwarz. Two academic institutions – ETH Zurich and Northeastern University – as well as simulation software specialist Ansys have also partnered with Nvidia.

These new partners offered up some suitably supportive statements backing Nvidia's efforts.

In addition to putting 6G near the top of its GTC agenda, Nvidia also had a big presence again at Mobile World Congress this year. It's a member of the newly-formed AI-RAN Alliance – which was announced on the eve of the show – and it also struck a broad AI partnership with Telenor. Nvidia's global head of business development for telco, Chris Penrose, also sat down with Telecoms.com to discuss telco AI.

Nvidia's telecoms strategy needed recalibrating after its bid to acquire chip designer Arm was blocked. Now, thanks to generative AI hype and its aim to be a telco AI enabler – not to mention its RAN acceleration hardware, and its bid to get in on the ground floor of 6G development – we are beginning to see the results of Nvidia's recalibrated strategy.