This post is also available in:
עברית (Hebrew)
As mobile networks become more complex, managing them is becoming a challenge in its own right. Future 6G systems are expected to support everything from autonomous vehicles and industrial automation to satellite connectivity and AI-driven applications. Traditional network architectures, which rely heavily on predefined rules and human operators, may struggle to keep pace with rapidly changing traffic demands and service requirements.
Researchers have now demonstrated a new type of 6G core network designed to manage itself. The system incorporates artificial intelligence directly into the network core, enabling it to learn, make decisions, and adjust operations automatically without requiring constant human intervention.
Unlike conventional 5G cores, where session management and routing structures remain relatively static, the new architecture continuously analyzes network conditions and optimizes how data flows through the system. AI algorithms can dynamically adjust routing paths, allocate resources, and modify quality-of-service parameters based on the needs of individual applications.
According to TechXplore, a key component of the design is the use of Segment Routing over IPv6 (SRv6), a technology that allows the network to create customized data paths for different services. Instead of treating all traffic the same way, the system can automatically assign optimized routes depending on factors such as latency, bandwidth requirements, or application type.
The researchers also developed an AI-native architecture that extends the service-based framework used in modern mobile networks. Supporting technologies include intelligent automation modules, reliability verification tools, and optimized AI training and inference capabilities. Together, these elements create a foundation for what researchers describe as an autonomous 6G core.
Testing showed a 40 percent improvement in session-processing efficiency compared with conventional architectures that rely on fixed routing paths. The system also demonstrated fine-grained control over network performance metrics, including bandwidth allocation and latency management.
One of the most notable achievements was the implementation of end-to-end AI automation using reinforcement learning. The network was able to recommend, select, and enforce operational policies autonomously, reaching what researchers describe as Level 3 automation, where session and traffic management occur without operator involvement.
From a defense and security perspective, self-managing communications networks could become increasingly valuable. Military operations, emergency response systems, and critical infrastructure often require resilient connectivity under changing conditions. Networks capable of automatically adapting to disruptions, congestion, or shifting mission requirements may improve both reliability and operational flexibility.
The technology has already been submitted for consideration within international telecommunications standardization efforts, highlighting the growing role AI is expected to play in the architecture of future 6G networks.
The research was published here.
