Deploying AI applications at the edge over a 5G connectivity fabric is an important step in driving digital transformation for enterprise operations. AI is reshaping how enterprises across every industry segment are doing their business.
5G offers superfast, wide-area, secure, low-latency and ultra-reliable connectivity, enabling enterprises to link their AI systems, whether on premises, in the field, or in the cloud. Together, these two forces will revolutionize industrial sectors and create value in enterprise markets. IHS Market expects the 5G-enabled value chain to generate $13.1 trillion of gross economic output by 2035.
Typically, 5G and edge AI infrastructures are evaluated, designed, procured, deployed, and managed separately. The siloed approach is inherently inefficient as AI and 5G run on computational power that can be provided by the same platform.
The first combined enterprise platform with edge AI and software-defined 5G
The NVIDIA AI-on-5G platform delivers an enterprise 5G network and relevant enterprise AI applications on a single computing platform. It is the first product in the market to do so in an end-to-end, fully orchestrated approach.
By bringing AI and 5G together on a single converged platform to deliver connected intelligence at the edge, the NVIDIA AI-on-5G platform fuels an array of low-latency enterprise AI applications, including precision robots, defect detection, video analytics, automated guided vehicles, digital twins, and more.
NVIDIA AI-on-5G brings AI and 5G together on one converged, GPU-accelerated platform, integrating 5G connected devices (including sensors and actuators). This makes it possible for enterprises to deploy edge AI applications over a private 5G network. The AI-first, cloud-native platform is based on NVIDIA-Certified™ Systems. It includes a hyperconverged, GPU-accelerated, edge “data center in a box” that can deliver combined high-performance computing for both AI workloads and software-defined 5G RAN. The same computing platform for AI workloads can be used for the 5G RAN, which is implemented as an additional software layer.
The NVIDIA AI-on-5G platform is an innovation in how intelligent edge networks are built, bringing flexibility, efficiency, reliability, and security to help redefine business processes and enable new use cases in manufacturing, retail, smart spaces, mining, and automotive sectors. This approach simplifies and accelerates enterprise adoption of AI and 5G, in a market that NVIDIA expects to create trillions in economic value.
The building blocks of AI-on-5G
AI-on-5G is made up of four building blocks:
- 5G Connected Devices: This includes IoT sensors (such as cameras) that capture inputs from the physical world and electro-mechanical actuators (such as robots) to execute actions in the physical world.
- 5G Radios: The O-RU is an O-RAN compliant 7-2 split radio unit to convert radio signals from the connected devices, through the antennas, to a digital signal. The NVIDIA AI-on-5G platform will support 5G radios from multiple OEMs. The O-RU is connected to the AI-on-5G platform using a time-sensitive fronthaul over eCPRI.
- AI-on-5G Hardware: This includes fronthaul termination and COTS servers that consist of the NVIDIA CONNECTX family of smart network interface cards (SmartNICs) and NVIDIAs latest converged accelerator. NVIDIA converged accelerator combines the powerful performance of the NVIDIA Ampere GPU architecture with the enhanced security and latency-reduction capabilities of the NVIDIA BlueField®-2 data processing unit (DPU).
- AI-on-5G Software: This includes NVIDIA Aerial SDK, 5G L2+ and 5G Core stack from partners, and Edge AI applications (from NVIDIA and partners). The entire unit is provisioned and managed with NVIDIA Fleet Command™ as illustrated in Figure 2.
NVIDIA is working with its partners to deliver commercial products based on the NVIDIA AI-on-5G platform. These partnerships are a significant feature in the telecommunications track at NVIDIA GTC, register to watch.
For more information on the AI-on-5G ecosystem and to subscribe for product updates, please visit the AI-on-5G webpage.