AI Transforming Radio Access Networks for 6G
Artificial intelligence is transforming Radio Access Networks (RAN) into GPU-accelerated, cloud-native platforms, according to a TelecomTV discussion featuring Nokia and NVIDIA. This evolution is seen as a way to enhance spectral efficiency and prepare network architectures for the demands of 6G.
- The Nokia and NVIDIA partnership involves a $1 billion investment by NVIDIA into Nokia to develop commercial-grade AI-RAN products. Future Nokia base stations will integrate NVIDIA's ARC platform, which combines Grace CPUs and Blackwell GPUs, to handle both RAN and AI workloads from the same infrastructure. - Standardization body 3GPP is laying the groundwork for AI in RAN through its 5G-Advanced specifications. Release 18 introduced studies on AI/ML for the air interface, including beam management and positioning, while Release 19 will specify a general framework for using AI/ML on the air interface. - The O-RAN Alliance is embedding AI into its architecture through components like the RAN Intelligent Controller (RIC) for tasks such as energy saving and MIMO optimization. The alliance is actively working with 3GPP to ensure its open and intelligent RAN solutions integrate into a unified 6G ecosystem. - At the international level, the International Telecommunication Union (ITU) has established a Focus Group on AI-Native for Telecommunication Networks (FG-AINN). This group aims to define the fundamental architectural shift required to embed AI into the core of the network, a key principle for 6G. - The development of 6G is a focal point in the technological rivalry between China and the United States, with control over standards seen as a way to gain economic leverage and ensure digital sovereignty. Europe is positioned as a key player in this competition, as it is home to major telecommunications equipment providers Nokia and Ericsson. - T-Mobile U.S. is collaborating with Nokia and NVIDIA to test and drive AI-RAN technologies as part of its 6G development. Field trials are expected to begin in 2026 to validate performance and efficiency gains in real-world environments. - The architectural goal is to create an "AI-native" RAN where the network is designed around AI from the ground up, rather than having AI capabilities added on. This approach, inspired by cloud-native principles, aims to enable automated network optimization, management, and new service delivery.
