Digital Twin Network-Based 6G Self-Evolution.

Area of Science:

• Telecommunications Engineering
• Artificial Intelligence
• Network Science

Background:

• Digital twins (DTs) are increasingly integrated into 6G network architectures for enhanced autonomy.
• Current DT applications primarily focus on network optimization, not self-evolution.

Purpose of the Study:

• To explore future directions for DT-native networks beyond mere autonomy.
• To introduce a novel architecture enabling network self-evolution through accurate predictions.

Main Methods:

• Proposed a DT-native architecture incorporating "future shots" for network predictions.
• Developed a long-term hierarchical convolutional graph attention model for cost-effective predictions.
• Utilized a conditional hierarchical graph neural network for strategy generation and efficient scale interactions.

Main Results:

• The proposed architecture enables accurate network predictions across various time scales.
• Facilitates the generation of effective self-evolution strategies for network elements.
• Demonstrates efficient interaction methods for small-to-large-scale network dynamics.

Conclusions:

• The "future shots" concept and proposed models advance DT-native networks towards self-evolution.
• This architecture is expected to facilitate high-level network autonomy in future 6G systems.
• Addresses the gap in achieving true network self-evolution using digital twins.