Agentic framework improves wireless network digital twin accuracy
Agentic Wireless Digital Twin Construction and Calibration Using Real-World Measurements
Information Theory
Summary
Wireless digital twins are like virtual models of real-world wireless networks that help in testing and improving network designs. The authors developed a new system that uses smart agents to build and fine-tune these models automatically using data from everyday smartphones. Their approach dramatically improves the accuracy of predicting signal strength by reducing errors. This method also works well with different devices by adjusting for small biases, making it a practical tool for large-scale wireless network simulations.
What this means in practice
- •For network planners: Create accurate virtual models of cellular networks to improve design and deployment using real user measurements without manual calibration.
- •For telecom operators: Use automatic digital twins to monitor and optimize network performance with minimal manual effort across different device types.$Commercial implications: Enables product offerings for automated network simulation and optimization targeting telecom operators seeking efficient network management.
Authors
Fan-Hao Lin, Kun-Li Shih, Chu-Hsiang Huang, Hui Chen, Chao-Kai Wen, Henk Wymeersch
Abstract
Wireless digital twins (WDTs) are promising enablers for developing and evaluating AI-native radio access networks, yet constructing a high-fidelity WDT typically requires substantial manual effort to integrate heterogeneous information and infer unknown propagation-related parameters. This paper proposes Agentic WDT (AWDT), an end-to-end agentic framework for autonomous WDT construction and calibration using readily available environmental information and measurements readily obtainable from commercial smartphones. AWDT comprises three agents: EnvAgent constructs the propagation environment, OpAgent infers BS and sector configurations, and MatAgent calibrates radio material properties. The agents iteratively refine the WDT using discrepancies between LTE/NR reference signal received power (RSRP) measurements and ray-tracing predictions. Real-world experiments show that AWDT reduces the RSRP prediction MAE from 11.46 to 5.25 dB, demonstrating a substantial improvement in ray-tracing fidelity. Evaluation with an independent measurement system further demonstrates cross-device transferability with lightweight device-specific bias adaptation, highlighting the potential of agentic AI for scalable WDT construction and calibration with limited prior knowledge.