SK Telecom and Japan's NTT Docomo have released a joint white paper outlining the direction for the virtualization of mobile base stations and the implementation of AI-based wireless networks. The document outlines the conditions for shifting the base station structure from one centered on dedicated equipment to one centered on software and general-purpose computing resources, and further enables the network to evolve into a structure capable of accommodating AI services.

SK Telecom announced on the 31st that it has jointly published a white paper with NTT Docomo outlining key technical requirements and development directions for the evolution of virtualized base stations (vRAN) and the implementation of AI-RAN. Based on both companies' experience in building and operating mobile networks, the white paper consists of an analysis of the technical elements required from the perspective of mobile carriers and the expected benefits.

The first condition presented in the white paper is the separation of hardware and software. The explanation is that if base station control software is designed not to be dependent on specific equipment or virtualization platforms, functions can be rapidly deployed or improved independently of infrastructure replacement. Both companies viewed this as a foundation for more flexibly expanding network functions and as a prerequisite for the advancement of virtualized base stations.

The second key element is resource pooling. By integrating distributed computing resources into a single pool, it is possible to simultaneously expand base station capacity and improve power efficiency while maintaining service quality. This approach aims to enhance network operational efficiency and utilization by transitioning base stations, which were previously operated as fixed equipment units, into a more flexible resource allocation system.

The third point is the adoption of AI computing capabilities. The white paper pointed out that in an xPU-based virtualized base station architecture that includes various computing resources such as CPUs, GPUs, and NPUs, orchestration technology that allocates and manages both AI and communication resources is crucial. It explained that through this, the virtualized base station can evolve into an integrated platform—namely, an AI-RAN—that processes not only communication functions but also AI computations.

The two companies unveiled demonstration results for resource pooling and xPU-based orchestration technologies at MWC 2026, held earlier this month. Furthermore, since signing a research cooperation agreement for next-generation communication infrastructure technologies in November 2022, they have published joint white papers on power saving technologies and 6G requirements in 2023, and on considerations for the deployment and operation of virtualized base stations in 2024. The two companies plan to continue their cooperation in the fields of strengthening 5G competitiveness and improving efficiency, as well as 6G standardization and technology verification.