NVIDIA Aerial Research Cloud Grows With New AI-Native Wireless Network Construction Tools
The AI Aerial platform’s portfolio of solutions enables developers to propel advancements in 6G and AI-RAN research.
The telecom sector is adopting AI more and more in order to maximise sustainability and spectral efficiency the quantity of data that can be sent per unit of bandwidth while providing smooth connections even in situations where signal strength is low.
The path towards AI-native wireless networks for 6G, which are constructed with AI and accelerated computing from the beginning to satisfy the demands of billions of AI-enabled connected devices, sensors, robotics, cameras, and autonomous cars, has been paved by advancements in AI-RAN technology.
Today, NVIDIA announced new tools in the NVIDIA Aerial Research Cloud portfolio to assist developers and telecom executives in pioneering these networks.
The Aerial Omniverse Digital Twin on NVIDIA DGX Cloud, the Aerial Commercial Test Bed on NVIDIA MGX, the NVIDIA Sionna 1.0 open-source library, and the Sionna Research Kit on NVIDIA Jetson are among the solutions in the extended portfolio that are assisting in the acceleration of AI-RAN and 6G research.
The NVIDIA Aerial Research Cloud portfolio is being used by more than 150 universities and research institutions in the US and abroad, such as Northeastern University, Rice University, The University of Texas at Austin, ETH Zurich, Fraunhofer Institute for Telecommunications, Heinrich-Hertz-Institut, HHI, Singapore University of Technology and Design, and the University of Oulu, to create, train, model, and deploy innovative AI-native wireless technologies. Leaders in the field like Amdocs, Ansys, Capgemini, DeepSig, Fujitsu, Key sight, Kyocera, MathWorks, Mediatek, Samsung Research, Soft Bank, and VIAVI Solutions are among these organisations.
Innovative Research and Development Instruments
For developers at every level of their research, from early experimentation to commercial implementation, the Aerial Research portfolio offers remarkable versatility and user-friendliness. Among its offerings are:
- NVIDIA DGX Cloud also offers Aerial Omniverse Digital Twin (AODT), a simulation environment for testing and optimising algorithms in physically accurate digital reproductions of complete wireless systems. Now, developers may access AODT from any location, including on-site, on laptops, in the public cloud, or through an NVIDIA cloud service.
- The full-stack AI-RAN deployment system known as the Aerial Commercial Test Bed (ARC-OTA) allows developers to test and deploy new AI models in real time over the air. It is currently available on NVIDIA MGX and can be purchased from manufacturers like Supermicro or as a managed solution through Sterling Skywave. To enable an end-to-end system for AI-RAN commercial testing, ARC-OTA combines commercial-grade Aerial CUDA-accelerated RAN software with open-source L2+ and 5G core from OpenAirInterface (OAI) and O-RAN-compliant 7.2 split open radio modules from WNC and LITEON Technology.
- With over 135,000 downloads, Sionna 1.0 is the most popular open-source library for communication systems research that is GPU-accelerated. The most recent version of Sionna has additional system-level simulation capabilities, a flexible link-level simulator, and a lightning-fast ray tracer for radio propagation.
- The NVIDIA Jetson platform powers the Sionna Research Kit, which combines a software-defined RAN based on OAI with accelerated computation for AI and machine learning applications. In just a few hours, researchers can use the kit to connect 5G equipment and start developing AI-RAN algorithms for next-generation wireless networks.
NVIDIA Aerial Research Cloud Environment for 6G and AI-RAN
The NVIDIA 6G Developer Program, an open community with over 2,000 members that represents top technology companies, academic institutions, research organisations, and telecom operators employing NVIDIA technologies to supplement their AI-RAN and 6G research, is part of the NVIDIA Aerial Research Cloud portfolio.
The development of AI-native wireless networks will heavily rely on testing and simulation. NVIDIA AODT is being used by companies like Amdocs, Ansys, Keysight, MathWorks, and VIAVI to improve their simulation solutions, and operators have built digital twins of their radio access networks to maximise performance in the face of shifting traffic conditions.
The NVIDIA Aerial Research Cloud portfolio was used to create nine of the ten demonstrations selected by the AI-RAN Alliance for Mobile World Congress, producing ground-breaking outcomes.
Using AI-based uplink channel interpolation, SoftBank and Fujitsu showed a throughput boost of up to 50% under unfavourable radio conditions.
Using neural networks, DeepSig created OmniPHY, an AI-native air interface that does away with conventional pilot overhead and can increase throughput by up to 70% in some situations. OmniPHY optimises spectral efficiency, lowers power consumption, and improves wireless network performance by incorporating machine learning into modulation, reception, and demodulation using the NVIDIA AI Aerial platform.
