NVIDIA’s AI-on-5G Platform Forms Centerpiece of Smart Everything

This article appeared in Microwaves & RF and has been published here with permission.

Tomorrow’s smart cities and factories, futuristic hospitals, and intelligent retail outlets will depend on both 5G and edge artificial intelligence (AI) computing. To that end, NVIDIA is teaming with partners such as Fujitsu, Google Cloud, Mavenir, Radisys, and Wind River to develop solutions for NVIDIA’s AI-on-5G platform.

Enterprises, mobile-network operators, and cloud-service providers that deploy the platform will be able to handle both 5G and edge AI computing in a single, converged platform. The AI-on-5G platform leverages the NVIDIA Aerial software development kit with the NVIDIA BlueField-2 A100—a converged card that combines GPUs and DPUs including NVIDIA’s “5T for 5G” solution. This creates high-performance 5G RAN and AI applications in an optimal platform to manage precision manufacturing robots, automated guided vehicles, drones, wireless cameras, self-checkout aisles, and hundreds of other transformational projects.

NVIDIA and several collaborators in this new AI-on-5G ecosystem are members of the O-RAN Alliance, which is developing standards for more intelligent, open, virtualized, and fully interoperable mobile networks. Such collaboration allows operators to use the same computing infrastructure required for 5G networking to provide AI services in enterprise, industrial, consumer, and residential settings.

The NVIDIA Aerial SDK, in combination with NVIDIA Metropolis, NVIDIA Isaac, and NVIDIA Clara, is an integral part of the AI-on-5G ecosystem and can be deployed on a single NVIDIA-Certified System using NVIDIA GPUs and DPUs on a single card.

A Range of Collaborations

As an example of NVIDIA’s collaborative efforts to enable AI-on-5G, Mavenir is building two 5G vRAN systems based on the Aerial SDK and will target the network operator segment for public 5G and for enterprise AI with private 5G. Mavenir and NVIDIA have created a hyperconverged enterprise 5G solution to enable enterprises to implement AI-on-5G applications in a seamless and easy-to-use solution.

For its part, Fujitsu later this year plans to deliver a 5G Open RAN system for verification starting with the sub-6-GHz band. Upon the system’s completion, Fujitsu and NVIDIA will be helping NTT DOCOMO and global operators with their evolution toward Open RAN in 5G, and beyond. Aerial software-defined implementation reduces time to market, speeds innovation and helps deliver AI applications to enterprises.

In the enterprise-deployment arena, Radisys and Wind River plan to deliver NVIDIA Aerial AI-on-5G solutions for enterprise 5G and industrial 5G networks. Over 6 million 5G cells will be deployed by 2027 to smart factories, fulfillment centers, and other enterprise, industrial, and public zones to provide localized connectivity solutions, according to ABI Research.

Google Cloud is extending the Anthos application platform to the network edge, allowing telecommunications service providers and enterprises to enable the rapid delivery of new services and applications at the 5G edge. Enterprises can turn to Google Cloud’s managed services and NVIDIA for their IoT economy, and to harness data and AI to drive business performance, improve operational efficiency, and optimize safety and reliability.

AI-on-5G Data Centers

Software-defined RANs are critical for building a modern 5G infrastructure that can run a range of applications on a common platform. NVIDIA Aerial enables the best possible utilization by providing elasticity as network traffic changes throughout the day and the flexibility to offer services based on dynamic customer needs.

The NVIDIA EGX platform brings AI computing capabilities to the edge where data gets created. The NVIDIA EGX stack, which is compatible with all commercially available Kubernetes infrastructures, is an ideal platform for Aerial enabling low-power, always-on, and high-performance devices, reshaping the telecom industry. Expanding on the platform, server makers also can pair NVIDIA GPUs and DPUs to build hyper-converged edge data centers.