NVIDIA and NTT DOCOMO revolutionize telecom services with world’s first GPU-accelerated 5G network

• The world’s first GPU-accelerated wireless integration marks a major leap in telecom services.
• This innovative 5G approach is poised to significantly cut costs and power use.

As the wave of generative AI pervades corporate boardrooms globally, worldwide telecommunications corporations are navigating ways to economically dispatch numerous novel AI applications to the edge via 5G and the forthcoming 6G networks.

By 2025, telecommunications companies aim to establish over 17 million 5G microcells and towers globally. The colossal task of constructing, managing, and refining this novel infrastructure, all the while upholding high-quality service delivery and enhancing customer satisfaction, stands as the imminent substantial hurdle for the industry.

Today marks a notable announcement from NTT DOCOMO, proclaiming the integration of a GPU-accelerated wireless solution within its network in Japan. This advancement distinguishes it as the pioneer global telco to incorporate a GPU-accelerated commercial 5G network.

Unlocking new potential in telecom services with Open RAN solutions

This strategic step by DOCOMO is directed towards confronting the significant issue of bolstering performance, reducing the total cost of ownership and enhancing energy efficiency, simultaneously unlocking the abundant potential for flexibility, scalability, and diversified supply chain offered by Open RAN.

The implemented 5G Open RAN solution employs a high-caliber 5G virtual radio access network (vRAN) from Fujitsu, grounded on the NVIDIA Aerial vRAN stack and NVIDIA Converged Accelerators. This amalgamation allows telecommunications companies to establish a thoroughly software- and cloud-defined network capable of dynamically allocating resources utilizing industry-standard equipment.

NTT DOCOMO’s Global Head of Open RAN solutions, Sadayuki Abeta, underscores the unparalleled flexibility and scalability offered by Open RAN in constructing next-generation 5G networks, a feat achievable thanks to multi-vendor connections.

This 5G Open RAN solution, being the foremost 5G vRAN for telco commercial deployment utilizing the NVIDIA Aerial platform, is paired with a robust, carrier-grade vRAN stack. The platform amalgamates the NVIDIA Aerial vRAN stack for 5G, AI frameworks, augmented compute infrastructure, and extended software support and maintenance.

Here’s more on NVIDIA Aerial platform:

In collaboration with Fujitsu and Wind River, the novel 5G solution capitalizes on the NVIDIA Aerial platform to curtail costs and diminish power consumption. As per DOCOMO, the solution slashes total costs by up to 30%, network design utilization by up to 50%, and base station power consumption by up to 50%.

Masaki Taniguchi, the Senior Vice President and Head of the Mobile System Business Unit at Fujitsu Limited, emphasized the significant achievement in delivering a 5G Open RAN network meeting the rigorous performance standards of operators. He highlighted the efficiency and high performance that network operators could achieve by employing the Fujitsu vCU/vDU in tandem with the NVIDIA Aerial platform.

NVIDIA’s engagement in the DOCOMO launch is part of a growing array of 5G solutions driving change in the telecommunications sector. Focusing on the NVIDIA Aerial vRAN stack and NVIDIA Converged Accelerators, along with NVIDIA BlueField data processing units (DPUs) and a suite of AI frameworks, NVIDIA provides high-performance, software-defined, cloud-native, AI-enabled 5G for on-site and telecommunications operators’ RAN.

Fujitsu, NVIDIA, and Wind River have cooperated under the umbrella of OREX (5G Open RAN service brand), inaugurated by DOCOMO in February 2021, for the development of the Open RAN 5G vRAN. OREX’s deployment in Japan is based on Fujitsu’s vDU and vCU, employing commercial off-the-shelf servers, the Wind River cloud platform, Fujitsu’s 5G vRAN software, and the NVIDIA Aerial vRAN stack and NVIDIA Converged Accelerators.

The role of NVIDIA Grace Hopper and NVIDIA BlueField DPU

The NVIDIA Grace Hopper Superchip fuses the NVIDIA Grace CPU, grounded on the Arm architecture, and the high-performance NVIDIA Hopper GPU. This integration, along with the BlueField DPU, contributes to achieving equivalent performance from full inline L1 offload akin to the NVIDIA Converged Accelerator.

Nevertheless, the pinnacle of performance enhancement stems from the integration of CPU and GPU architectures utilizing NVIDIA NVLink-C2C, delivering a harmonious CPU+GPU memory model for accelerated tasks like 5G vRAN.

NVIDIA NVLink-C2C, the memory-coherent, high-bandwidth, and low-latency interconnect from NVIDIA, delivers a staggering up to 900 GB/s total bandwidth, a leap that is seven times higher than the bandwidth of the commonly employed x16 PCIe Gen5 lanes in accelerated systems.

OREX’s expansion from Japan and beyond

The commercial deployment of a 5G Open RAN network by NTT DOCOMO, utilizing the NVIDIA 5G platform, stands as a landmark for the telecommunications sector, demonstrating the prowess of GPU-based acceleration for computationally intensive L1 PHY processing.

This novel network delivers augmented performance, flexibility, and scalability, along with enhanced cell density, substantial advancements in energy efficiency, and a slash in TCO, paving the path for the broad-based adoption of GPU-based acceleration in cellular RANs.

In unison with OREX partners, DOCOMO is ardently working to endorse a multi-vendor, Open RAN–compliant 5G vRAN to the global operator community. The commercial deployment in Japan aligns with OREX’s vision, allowing its members to commercially validate their solutions and subsequently promote them to other global operators.

NVIDIA persistently collaborates with DOCOMO and additional partners, extending support to operators globally in the deployment of high-performance, energy-efficient, software-defined, commercial 5G vRAN.

---

---

---