In traditional mobile communication networks, every video call or short video refresh requires the network to "manually" establish a dedicated service channel. Now, a research team from the Electronics and Telecommunications Research Institute (ETRI) in South Korea has enabled AI to take over this process—the network no longer requires human intervention, with AI autonomously deciding how to establish sessions, allocate resources, and schedule paths.

In June 2026, according to the latest issue of IEEE Communications Magazine, the ETRI research team successfully developed key technologies for an AI-driven 6G core network, achieving the "Intelligent 6G Core Network" for the first time in South Korea. This technology improves session processing efficiency by approximately 40% compared to traditional architectures, marking a new era where next-generation mobile communication networks transition from "static configuration" to "autonomous intelligence."

The "Static Dilemma" of Traditional Core Networks

In mobile communication networks, a "session" refers to a "service channel" established by the network for a specific service—whether making a video call, watching short videos, playing games, or downloading files, the network must establish and manage sessions for these services, including allocating resources, selecting transmission paths, and ensuring latency and bandwidth.

However, traditional 5G core networks rely on relatively fixed session management structures, with data flow primarily controlled by predefined rules. This "static" architecture, in the 6G era where service types are increasingly diverse and quality of service requirements are more stringent, is exposing pain points such as insufficient flexibility, low resource utilization, and high operational costs.

How to make the core network "dynamic," capable of real-time self-adjustment based on vastly different service demands, has become a core challenge in 6G technology development.

Three Pillars of AI "Taking Over" the Network

The ETRI research team achieved systematic innovation at three levels—architecture, path, and strategy—installing an "AI brain" into the 6G core network.

Architectural Change: AI-Native Control Plane and User Plane

The team developed an AI-native control plane and user plane architecture, breaking the relatively isolated design paradigm of control and user planes in traditional 5G core networks. This architecture deeply embeds AI into the network's control logic, enabling the network to continuously learn, make autonomous decisions, and complete operational management with minimal human intervention.

Path Change: SRv6 Enables "On-Demand Routing" for Data

The research team integrated Segment Routing over IPv6 (SRv6) technology into the system, allowing the network to automatically configure data transmission paths based on service characteristics and dynamically establish and adjust sessions and routes. Compared to traditional GTP architectures that rely on fixed transmission paths, this new technology achieves differentiated routing for different application scenarios.

Strategy Change: Reinforcement Learning Drives "Level 3 AI Automation"

The team introduced a reinforcement learning-based policy recommendation mechanism into the network control system, achieving Level 3 end-to-end AI automation. This means session management and traffic control can be completed automatically without human involvement, with AI autonomously formulating and executing control policies across the entire network.

Level 3 AI automation refers to the evolution level of the network from human-assisted decision-making to fully autonomous operation. Reaching Level 3 means AI possesses full autonomous decision-making and execution capabilities, capable of completing session and traffic control without any human intervention.

The "Hard Metrics" Behind the 40% Efficiency Improvement

According to official ETRI announcements, performance evaluation results show that compared to traditional fixed routing paths based on GTP architecture, the new system improves session processing efficiency by approximately 40%.

In addition to efficiency gains, the system achieves fine-grained control over quality of service (QoS) parameters such as latency and bandwidth, enabling differentiated service guarantees for different business needs.

Currently, this architecture has been submitted to the 3GPP SA2 standardization working group for discussion. Over 60 core patents have been filed related to this technology, and research findings have been published in top academic journals such as IEEE Communications Magazine.

Paving the "Intelligent Foundation" for 6G Commercialization

Significantly Reducing Operator Operational Costs

Through AI-driven automation, human intervention in network operations is minimized. This is of great significance for telecom operators facing exponential growth in network complexity and rising operational costs. AI autonomous decision-making means fewer operations personnel, faster fault response, and higher resource utilization.

Supporting Diverse 6G Service Scenarios

The 6G era will see a surge in differentiated application scenarios—from immersive XR to holographic communications, from industrial internet to low-earth orbit satellite communications. This technology achieves differentiated routing and quality of service guarantees for different business needs, providing core network support for the 6G vision of "service programmability."

Paving the Way for Integration with Low-Earth Orbit Satellite Communications

The research team's next step is to develop core network and transport network integration technologies supporting low-earth orbit satellite communications, building a new-generation 6G core architecture that integrates services, computing, and networking. This means future 6G networks will not only serve terrestrial users but also establish integrated space-terrestrial communication links.

The "Korean Solution" in Global Competition

Taesik Cheung, Director of the ETRI Network Research Division, stated: "This work is an important milestone, demonstrating that the 6G core network is evolving beyond a simple data processing system into an intelligent platform for AI autonomous decision-making and control."

This research was funded by the "6G Core Technology Development Program" of the Ministry of Science and ICT (MSIT) and the Institute for Information & Communications Technology Planning & Evaluation (IITP) in South Korea, and was jointly advanced with major Korean telecom companies such as SK Telecom, LG Uplus, and SNETICT.

From technological breakthroughs to patent portfolios, from academic publications to international standard promotion, the ETRI team has built a complete innovation chain. As the global 6G standard competition enters a critical phase, this AI core network technology, boasting a "40% improvement in session processing efficiency," is becoming a key card for South Korea on the 6G track.