During the development of the Sixth Generation (6G) networks, the integration of Artificial Intelligence (AI) into network systems has become a focal point, leading to the concept of AI-native networks. High quality data is essential for developing such networks. Although some studies have explored data collection and analysis in 6G networks, significant challenges remain, particularly in real-time data acquisition and processing. This paper proposes a comprehensive data collection method that operates in parallel with bitstream processing for wireless communication networks. By deploying data probes, the system captures real-time network and system status data in software-defined wireless communication networks. Furthermore, a data support system is implemented to integrate heterogeneous data and provide automatic support for AI model training and decision making. Finally, a 6G communication testbed using OpenAirInterface5G and Open5GS is built on Kubernetes, as well as the system's functionality is demonstrated via a network traffic prediction case study.
@article{arxiv.2509.01276,
title = {A Real-time Data Collection Approach for 6G AI-native Networks},
author = {He Shiwen and Dong Haolei and Wang Liangpeng and An Zhenyu},
journal= {arXiv preprint arXiv:2509.01276},
year = {2025}
}