LITEON and NVIDIA Partner on AI-RAN
At MWC Barcelona, LITEON Technology announced it's integrating its Open Radio Units with NVIDIA's AI Aerial platform. The collaboration is designed to speed up the commercialization of AI-native, GPU-accelerated radio access network (RAN) architectures.
This collaboration transforms the Radio Access Network (RAN) from a hardware-defined system into a software-defined, AI-native platform. By running RAN and AI workloads on the same NVIDIA GPU-powered infrastructure, telecom operators can dynamically allocate resources, potentially doubling or tripling capacity utilization compared to using separate hardware. LITEON provides the critical hardware component: the O-RAN-compliant 5G Open Radio Unit (O-RU). This piece of the puzzle ensures its radio equipment can seamlessly integrate with NVIDIA's compute and software, a key tenet of the Open RAN model which aims to create a multi-vendor, interoperable network architecture. NVIDIA's Aerial platform is a software development kit for building high-performance, GPU-accelerated 5G and 6G applications. It includes the Aerial CUDA-Accelerated RAN, which uses NVIDIA's massive parallel processing capabilities for intense signal processing tasks, and a digital twin platform using Omniverse for simulating entire wireless networks. The "AI-native" aspect means embedding AI/ML models directly into the network's signal processing layers. This allows for real-time network optimization, predictive maintenance, and enhanced spectral efficiency, which is crucial as networks prepare for the leap from 5G to 6G. For robotics and autonomous systems, this architecture is a game-changer. The massive bandwidth and ultra-low latency of an AI-RAN are essential for teleoperations of robots, autonomous vehicles, and real-time computer vision applications at the edge. Processing AI workloads directly on the network infrastructure reduces the round-trip time for critical decisions. LITEON is also working with partners like SynaXG and Supermicro to create a unified system. This ecosystem approach, built on open standards, aims to reduce integration complexity and speed up the deployment of commercial AI-RANs for use cases in smart factories, IoT, and smart cities.
