Nvidia is preparing to launch a cloud-based research platform for 6G R&D projects. The Nvidia 6G Research Cloud provides access to RAN AI models with potential to accelerate 6G project pipelines, the chipmaker said. 

It consists of three “foundational” elements. First, Nvidia’s Aerial Omniverse Digital Twin for 6G is a reference model containing virtual simulations of complete 6G towers and city networks. The toolkit incorporates software-defined RAN and user-equipment simulators, along with “realistic” terrain and object properties. 

Next, the company is providing a software-defined RAN stack for programming and testing 6G networks. The Nvidia Aerial CUDA (Compute Unified Architecture)-Accelerated RAN toolkit enables “significant flexibility” within a real time workflow for customised 6G access networks.

Finally, Nvidia said 6G research would be supported by its Sionna Neural Radio Framework, which enables Nvidia GPUs to be leveraged for at-scale AI training alongside common software frameworks for machine learning code, such as PyTorch and TensorFlow. 

Nvidia is also making available its “link-level research tool” for AI and machine learning-based wireless simulations. The Nvidia Sionna open-source library enables rapid communication system prototypes with native support for machine learning in 6G signal processors.

Ansys, Arm, ETH Zurich, Fujitsu, Keysight, Nokia, Northeastern University, Rohde & Schwarz, Samsung, SoftBank and Viavi are among the first adopters and partners for the 6G research platform. To access the platform, researchers can sign up for the Nvidia 6G Developer Program.

El artículo original puede consultarse en: