The nexus of AI, the Internet of Things (IoT), and accelerated computing at the edge is turbocharging network performance and efficiency. The world’s leading telcos are turning to NVIDIA technologies to create AI-enabled solutions, build software-defined infrastructure for 5G, and bring connected intelligence to smart devices at the edge.
View Report: State of AI in Telecom 2023
Don't miss these three upcoming telecom sessions at GTC.
Fujitsu and the NTT Docomo-led 5G Open RAN Ecosystem (OREC) have delivered a 5G OpenRAN solution based on NVIDIA platform with the most competitive total cost of ownership in the market. The solution is a high performance, ultra-efficient, 100% software-defined and 6G-ready 5G OpenRAN. It relies on one architecture to address lowband, midband, and mmwave, making it extensible and 6G-ready. It uses standard 2U DU node to support 100MHz 4T4R * 32 cells based on a 50% PRB load, providing excellent energy efficiency and a significant improvement in spectral efficiency. Thanks to being fully software-defined, AI-enabled, and cloud-native, it supports massive MIMO, RAN sharing, and the dynamic scaling of RAN and AI workloads in the cloud. We'll discuss the technical details of the solution and how the partnership between Docomo/OREC, Fujitsu, and NVIDIA paved the way for deployment in commercial public telecom networks.
NTT has always been an innovator, and is now at the forefront of defining what multiaccess edge computing (MEC) should be. Learn from NTT DOCOMO, NTT EAST, and NTT Communications how different approaches to MEC come to fruition to enable AI enterprise applications. NTT DOCOMO has launched a new nationwide 5G MEC design for GPU-accelerated enterprise applications like IVA and graphics/XR that requires 5G high throughput/low latency. The solution includes an edge stack designed for high-performance applications based on NVIDIA GPUs, smart NICs, and switches. Learn more about the essential technologies of NTT Docomo 5G MEC to deliver accelerated applications.
Fraud detection is a challenging task since fraudsters continuously change their behavior, they may represent rare cases, and fraud patterns may even be unseen during training. Expert human inspectors develop an intuition of possible fraud[1]ulent cases. Still, they need an accurate and automated process to help them spot fraud out of thousands of daily samples. Fraud detection is treated as a binary classification problem where the positive class is of extreme interest and may be highly underrepresented in the data. In this talk, we evaluate XGBoost on balanced and imbalanced data, and present key points that helped us improve our models in terms of detection precision, recall, and training speed. We will explain the concepts behind our choices and present code examples in well-known python libraries. Finally, we will reflect on lessons learned from our experience using XGBoost for fraud detection, considering that the types of mistakes a binary classifier makes have different impacts on different applications.
Telecoms giant adopts NVIDIA AI to process data more efficiently, optimize service-vehicle routing, and create digital twins and avatars.
A new AI-on-5G solution enables telcos to deliver immersive graphics, metaverse applications, and computer vision from one GPU-accelerated system.
Four hundred respondents from the world’s leading telecommunications and communication service providers suggest AI is important to revolutionize the industry, but change appears slow-going.
Enterprises across industries are adding business value by adopting edge AI technologies, including computer vision, extended reality, and robotics. Telcos, with connectivity and edge infrastructures, are uniquely positioned to provide these services.
What is clear is that 5G is set to shift operators’ service models away from monthly voice, data, and messaging bundles and towards higher-value services that are tuned for specific customer requirements—and that it will increasingly make use of edge computing.
-"Developing an Edge in 5G Services"
Network operations are becoming increasingly complex, and vast amounts of valuable data are being generated. By leveraging AI to extract timely, actionable insights from this data, telcos can optimize network operations, improve customer experiences, and unlock new revenue streams.
T-Mobile deployed a conversational AI-based chatbot, a self-service center, agent-assist functionality, and transcription services in their customer experience centers. Conversations between customers and agents are now transcribed to text—live—to better assist customer experience center agents. According to T-Mobile, in the first 18 months, they realized a significant return on their investment.
NVIDIA networking telco solutions provide intelligent offloads and accelerators to maintain adherence to stringent Radio Access Network (RAN) timing specifications, offer real-time edge AI processing, and provide an accelerated and efficient infrastructure for 5G and network services applications.
NVIDIA Accelerated Switch and Packet Processing (ASAP2) offloads the virtual switch data plane to the SmartNIC or DPU, freeing precious CPU resources. Combining the performance and efficiency of server/storage networking hardware with the flexibility of virtual switching software provides software-defined networks with the highest total infrastructure efficiency, deployment flexibility, and operational simplicity.
Data Plane Development Kit (DPDK) is a programming framework to provide faster packet processing and deliver increased performance and efficiency for applications requiring high-speed networking access. NVIDIA provides an input/output (I/O) packet performance breakthrough that eliminates the performance barrier associated with deploying network function virtualization (NFV) solutions in both service provider networks and enterprise cloud deployments.
NVIDIA time-triggered transmission technology for telco, referred to as 5T-for-5G technology, provides time synchronization to achieve timing accuracy for network traffic and is a real-time and high-performance solution for building an efficient, time-synchronized Cloud RAN infrastructure.
NVIDIA Aerial™ is pushing the limits of wireless innovation for 5G RAN. Aerial delivers a high-performance, software-defined, programmable, GPU-accelerated, and ORAN-compliant 5G virtual RAN (vRAN). Aerial is a key building block for telco 5G networks, AI-on-5G for private 5G, and RAN-in-the-cloud solutions. NVIDIA is working with partners on ongoing research, development, and ecosystem engagement for 6G.
While the number of RAN sites is projected to nearly double to over 17 million by 2027(1), the reality is that many have an average utilization below 25 percent(2). This means that the strategic importance of the RAN presents a huge opportunity for new ideas on how to plan, deploy, manage, and monetize the RAN. Both AI and accelerated computing have a huge role to play.
Explore the use cases driving telco investment in AI, including examples of telco deployments, and get an introduction to the ecosystem that can support your AI journey.
Learn about telcos who are working with NVIDIA’s AI ecosystem and the applications they’re investing in to capitalize on enterprise service opportunities.
Learn why a DPU or SmartNIC is one of the best ways to improve data center efficiency and reduce server power consumption by up to 30 percent.
Explore advances in 5G network acceleration and security, how AI-on-5G and edge applications are delivering new services, and how telcos are improving operations with AI.
Stay tuned for telco news and content, delivered straight to your inbox.
Sign up for enterprise news, announcements, and more from NVIDIA.