AI news ·
ETRI develops autonomous 6G core network powered by artificial intelligence
ETRI built Korea's first AI-powered 6G core, replacing static management with autonomous control. This boosts processing efficiency by 40% to reach Level 3 automation.
The Electronics and Telecommunications Research Institute (ETRI) has developed an AI-powered intelligent core for 6G networks, marking the first realization of this technology in Korea. By replacing static session management with autonomous AI control, the system improves session processing efficiency by 40% compared to conventional fifth-generation architectures.
How the intelligent 6G core operates
Unlike conventional 5G core networks that rely on fixed routing paths, this new architecture uses AI to optimize service-specific sessions, routing paths, and quality of service in real time. The network incorporates Segment Routing over IPv6 (SRv6) to dynamically configure data transmission paths based on service characteristics. This setup allows for the automatic establishment and adjustment of sessions, ensuring performance-optimized connectivity without manual operator intervention.
Core components and automation capabilities
To achieve this, the research team built three main components: an AI-native control and user plane architecture extending the Service-Based Architecture (SBA), a dedicated module for reliability verification and AI Agents & Automation, and training optimization technologies for AI applications. Performance evaluations show a 40% improvement in session processing efficiency compared to conventional GTP-based architectures. By integrating reinforcement learning-based policy recommendations, the solution reaches Level 3 End-to-End AI automation, allowing the system to determine and enforce policies across the entire network independently.
Standardization efforts and next steps
ETRI has submitted the proposed architecture to the 3GPP SA2 standardization group, where it remains under active discussion. The institute has also filed more than 60 core patents and published related findings in IEEE Communications, demonstrating progress across technical innovation, standardization, and academic AI for Science & Research domains. The project was conducted under the 6G Core Technology Development Program, supported by the Ministry of Science and ICT, the Institute of Information & Communications Technology Planning & Evaluation, SK Telecom, LG Uplus, and SNETICT.
Taesik Cheung, assistant vice president of the Network Research Division, said the development proves that 6G core networks are evolving beyond simple data processing systems. "This work represents a major milestone, demonstrating that 6G core networks are evolving beyond simple data processing systems into intelligent platforms where AI autonomously makes decisions and controls operations," Cheung said. "We will continue to advance next-generation 6G core technologies based on the evolved SBA framework." Namseok Ko, director of the Mobile Core Network Research Section, added that integrating AI-driven control with network automation establishes a practical foundation for realizing AI-native 6G.
Why this matters for science and research professionals
Researchers and engineers working on next-generation telecommunications will see how autonomous AI shifts from theoretical models to deployed infrastructure. The 40% efficiency gain and Level 3 automation provide measurable benchmarks for evaluating future AI-native network designs. Furthermore, the submission to 3GPP SA2 means these architectural choices will directly influence global 6G standards, making early familiarity with SBA extensions and reinforcement learning policy mechanisms a practical necessity for network architects.