NVIDIA's Cellular Software-Defined Radio (SDR) solutions are designed to transform traditional cellular networks into flexible, software-driven infrastructures capable of supporting advanced 5G and future 6G technologies. By leveraging GPU acceleration and AI integration, these solutions enable efficient, scalable, and programmable radio access networks (RANs).
Key Components:
NVIDIA Aerial Platform
This platform provides CUDA-accelerated RAN libraries for Layer 1 (L1) and Layer 2 (L2) processing, allowing complex computations to be executed more efficiently than on non-GPU solutions. It supports dynamic allocation of 5G and AI workloads within the same GPU, enhancing resource utilization and energy efficiency.
AI Integration
NVIDIA's AI Aerial platform incorporates AI algorithms to improve spectral efficiency and network performance. For instance, advanced channel estimation techniques, such as the Reproducing Kernel Hilbert Space (RKHS) method, are employed to enhance signal quality and user experience.
Aerial Research Cloud
This fully programmable 5G and 6G network research sandbox enables researchers to rapidly simulate, prototype, and benchmark innovative new software deployed through over-the-air networks. It democratizes 6G innovations with a full-stack, C-programmable 5G network, and jumpstarts machine learning in advanced wireless communications using NVIDIA-accelerated compute
Product Document
- Aerial CUDA-Accelerated RAN
- Getting Started with pyAerial
- NVIDIA Sionna: An Open-Source Library for 6G Physical-Layer Research
- Sionna/Github
Papers/Whitepapers
- X5G: An Open, Programmable, Multi-vendor, End-to-end, Private 5G O-RAN Testbed with NVIDIA ARC and OpenAirInterface - arXiv (2024)
- Low-Latency CUDA LDPC Decoder for SDR Solutions
- Deepwave Digital Creates an AI Enabled GPU Receiver for a Critical 5G Sensor
- NVIDIA SDR (Software Defined Radio) Technology
- Simplifying AI for Communications, Radar, and Wireless Systems - Deepwave Digital
- Real-Time Inference of 5G NR Multi-user MIMO Neural Receivers - Github
General Reading
- NVIDIA-Accelerated Radio Access Networks (RANs)
- Transform Your Cellular Network With AI-RAN - NVIDIA
- Data Plane Development Kit (DPDK*) - Intel
- Nvidia Pegs Software-Defined Radios, Edge Computing as Key to 5G Cellular - Lime microsystems (2019)
- Boosting Inline Packet Processing Using DPDK and GPUdev with GPUs - NVIDIA TechBlog
- Introducing NVIDIA Aerial Research Cloud for Innovations in 5G and 6G - NVIDIA (2023)
- Enhanced DU Performance and Workload Consolidation for 5G/6G with NVIDIA Aerial CUDA-Accelerated RAN - NVIDIA (2024)
- T-Mobile, NVIDIA, Ericsson, and Nokia: Transforming Mobile Networks with AI-RAN - 5G World Pro (2024)
YouTube
- 1.5: DPDK Introduction - FD.io (2016)
- Efficiently Deploying GPU Accelerated 5G CloudRAN for Edge AI Inferencing - NVIDIA Developer (2020)
- Data Plane Development Kit (DPDK ) - TechUpskill (2021)
- Introduction to DPDK - Root Access (2022)
- DPDK in real-time GPU packet processing applications - Elena Agostini, NVIDIA - DPDK Project (2022)
- Real-time and low latency media transport stack based on DPDK - Ping Yu & Frank Du, Intel - DPDK Project (2022)
- Exploring DPDK's Role in 5G Architecture - DPDK Project (2023)
- Introduction to NVIDIA DOCA and DPU programming - Elena Agostini - NVIDIA - HiCrest (2023)
- 5G RAN Acceleration with GPU and DPDK - Elena Agostini, NVIDIA - DPDK Project (2023)
- 5G UPF Acceleration with DPDK - Gal Cohen, NVIDIA - DPDK Project (2023)
- AI for 5G Advanced toward 6G - Oregon ComSoc (2023)
- NVIDIA 6G Research Cloud Platform - NVIDIA Developer (2024)
- Future of Cellular: T-Mobile & NVIDIA Accelerate AI-RAN Commercialization - NVIDIA (2024)