NVIDIA AI Aerial Optimizes Wireless Networks, Delivers Next-Generation AI Experiences on One Platform
With an AI computing infrastructure, telecommunications companies are moving beyond voice and data services to optimize wireless networks and meet the demands of generative AI on mobile, robotics, autonomous vehicles, smart factories, 5G, and many other areas.
A set of accelerated computing hardware and software called NVIDIA AI Aerial was unveiled today with the goal of developing, modeling, training, and implementing AI radio access network technology (AI-RAN) for wireless networks in the AI age.
The platform will develop into an essential building block that enables large-scale network optimization to meet the needs of numerous new services. As a result, there will be large total cost of ownership savings and new income streams for enterprise and consumer services for telecom providers.
Telecommunications service providers can now support generative AI-driven co-pilots and personal assistants, teleoperations for manufacturing robots and autonomous vehicles, computer vision in manufacturing and agriculture, logistics, emerging spatial computing applications, robotic surgery, 3D collaboration, and 5G and 6G advancements thanks to NVIDIA AI Aerial.
AI-RAN
Driving Future Networks With AI-RAN
The first AI-RAN platform in the world, NVIDIA AI Aerial, can host generative AI, manage RAN traffic, and incorporate AI into network optimization.
With edge AI apps to host internal and external generative AI applications, AI-RAN provides software-defined RAN that is both high-performance and energy-efficient. It also improves network experience and opens up new revenue streams.
The multifunctional networks of the future that depend on AI-powered telecommunications capabilities are built on AI-RAN.
Using NVIDIA AI Aerial in the Telecom Sector
In order to enable telecom operators to engage at any point from development to deployment for next-generation wireless networks, the NVIDIA AI Aerial platform provides access to a full range of capabilities, including a high-performance, software-defined RAN along with training, modeling, and inference options.
Among the features of the NVIDIA AI Aerial platform are:
- Software libraries are included in NVIDIA Aerial CUDA-Accelerated RAN to help partners create and implement high-performance virtualized RAN workloads on computing platforms that are accelerated by NVIDIA.
- The PyTorch and TensorFlow software libraries included in the NVIDIA Aerial AI Radio Frameworks are used to create and train models that enhance spectral efficiency and introduce new functionalities to the processing of 5G and 6G radio signals. NVIDIA Sionna, a link-level simulator that facilitates the creation and training of neural network-based 5G and 6G radio algorithms, is also included in this.
- A framework for developing network digital twins at the system level is called NVIDIA Aerial Omniverse Digital Twin (AODT). With the use of AODT, wireless networks can be simulated with physical accuracy, ranging from a single base station to a vast network with numerous base stations spanning a whole city. It includes realistic terrain and object attributes of the actual world, user-equipment simulators, and software-defined RAN (Aerial-CUDA Accelerated RAN).
NVIDIA Innovation Center for AI Aerial and AI RAN
With the launch of the AI-RAN Innovation Center, NVIDIA is working with T-Mobile, Ericsson, and Nokia to quicken the commercialization of AI-RAN.
The facility will make use of the NVIDIA AI Aerial platform’s primary features. Through the development of AI-RAN, the partnership aims to bring RAN and AI innovation closer together to give customers’ revolutionary network experiences.
Ericsson’s investment in its AI-RAN technology, communications service providers may now implement portable RAN software that works on a variety of platforms.
The NVIDIA AI Aerial Environment
Softbank and Fujitsu are important members of the NVIDIA AI Aerial ecosystem.
For testing and simulation purposes, Ansys and Keysight use the NVIDIA Aerial Omniverse Digital Twin, and academic partners including Deepsig, ETH-Zurich, Northeastern University, and Samsung work together on 6G research and NVIDIA Aerial AI Radio Frameworks.
Key partners for NVIDIA AI Aerial include cloud stack software companies like Aarna Networks, Canonical, Red Hat, and Wind River; networking stack providers like Arrcus network; and server infrastructure providers like Dell Technologies, Hewlett Packard Enterprise, and Supermicro. AI solution decision-making is speeding up with the help of edge solution providers like Vapor.io and system integrators like World Wide Technology and its AI Proving Ground.
