An eXtended Reality Offloading IP Traffic Dataset

本数据集由诺基亚贝尔实验室西班牙分部创建，专注于扩展现实（XR）技术中的数据卸载场景，特别是针对高要求的XR卸载场景。数据集通过一个完整的端到端XR卸载解决方案捕获，旨在为学术界和工业界提供设计和测试XR卸载解决方案的数据和工具。数据集包括多种分辨率的上行和下行流，用于两种不同的卸载场景：全XR卸载和以自我为中心的人体分割算法卸载。此外，数据集还包含从捕获数据中获得的XR流量模型，以及用于生成合成数据的简单一致方法。通过使用开源5G无线接入网络（RAN）模拟器，验证了模型的准确性，展示了数据集在设计和优化无线网络解决方案中的应用价值。

This dataset was developed by Nokia Bell Labs Spain, focusing on data offloading scenarios in extended reality (XR) technologies, particularly for high-demanding XR offloading use cases. Captured via a complete end-to-end XR offloading solution, it aims to provide academia and industry with data and tools for designing and testing XR offloading solutions. The dataset includes uplink and downlink streams of various resolutions for two distinct offloading scenarios: full XR offloading and ego-centric human segmentation algorithm offloading. Additionally, it contains XR traffic models derived from the captured data, as well as a simple and consistent methodology for generating synthetic data. The accuracy of these models was validated using an open-source 5G Radio Access Network (RAN) simulator, demonstrating the dataset's practical value in designing and optimizing wireless network solutions.