基于环境图的信道多径预测研究

张兆灵; 金婧; 赵婧博; 于力; 蔡逸辰; 马良; 张建华

doi:10.11999/JEIT260416

基于环境图的信道多径预测研究

doi: 10.11999/JEIT260416 cstr: 32379.14.JEIT260416

张兆灵^1,,
金婧^2, ,,
赵婧博²,
于力¹,
蔡逸辰¹,
马良²,
张建华¹

1.
北京邮电大学网络与交换技术全国重点实验室北京 100876
2.
中国移动通信有限公司研究院北京 100053

基金项目: 国家自然科学基金项目(No. 62525101, No. 62401084)，国家重点研发计划项目(No. 2023YFB2904805)，受北京邮电大学-中国移动联合研究院资助

详细信息

作者简介:
张兆灵：女，学生，研究方向为信道预测，无线信道知识，邮箱　zhangzhaoling@bupt.edu.cn

金婧：女，高级工程师，研究方向为数字孪生，通信感知一体化，邮箱　jinjing@chinamobile.com

赵婧博：男，工程师，研究方向为数字孪生，通信感知一体化

于力：男，副研究员，研究方向为信道建模和预测，机器学习，环境感知和重构，信道数字孪生等

蔡逸辰：女，学生，研究方向为信道预测

马良：男，工程师，研究方向为数字孪生，通信感知一体化

张建华：女，北京邮电大学教授，博士生导师，网络与交换技术全国重点实验室副主任，北邮-移动联合研究院执行院长，研究方向为6G移动通信技术，人工智能，数据挖掘，智能信道建模，大规模MIMO和太赫兹信道建模，信道仿真仪，OTA测试等

通讯作者:
金婧　jinjing@chinamobile.com

中图分类号: XXXXXXXXX
计量
- 文章访问数: 17
- HTML全文浏览量: 4
- PDF下载量: 0
- 被引次数: 0
出版历程
- 修回日期: 2026-08-10
- 录用日期: 2026-08-10
- 网络出版日期: 2026-08-15

Research on Channel Multipath Prediction Based on Environmental Graph

1.
State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications, Beijing 100876, China
2.
China Mobile Research Institute, Beijing 100053, China

Funds: National Natural Science Foundation of China under Grant Numbers 62525101 and 62401084, National Key Research and Development Program of China under Grant Number 2023YFB2904805, Beijing University of Posts and Telecommunications - China Mobile Communications Group Co.,Ltd. Joint Institute

摘要

摘要: 面向6G智能空口、智能化网络及AI与通信的深度融合需求，如何从结构化环境信息中提取与传播路径相关的知识，是构建环境驱动信道模型的重要问题。针对传统信道建模方法难以刻画发射端(Transmitter, Tx)、接收端(Receiver, Rx)与散射体间空间结构及交互关系的局限性，本文提出融合信号传播机制的环境图结构建模方法，将Tx、Rx及散射体抽象为图节点，根据节点间的空间距离与可视性构建边特征，并基于边增强图同构网络(Edge-aware Graph Isomorphism Network, EGIN)学习环境图中的结构关联，实现参与传播的有效散射体检测；进一步根据候选散射体构建单跳与双跳候选路径，设计路径特征学习网络与有效性排序模型，实现候选传播路径预测。本文在单一工业物联网室内仿真场景的3381个Rx位置上进行验证，结果表明所提方法能够在该场景内提取与传播路径相关的结构特征，提高有效散射体检测与候选路径排序的准确性。
- 环境图结构 /
- 图神经网络 /
- 有效散射体 /
- 传播路径预测
Abstract: Objective Environment-aware channel modeling requires a structured representation that can connect physical objects in a propagation environment with the resulting multipath topology. However, conventional data-driven methods generally treat environmental information as unstructured global features and therefore have difficulty representing the interactions among the transmitter (Tx), receiver (Rx), and surrounding scatterers. This study investigates whether an environmental graph can provide an effective intermediate representation for identifying propagation-relevant scatterers and predicting candidate multipath structures. Methods An environmental graph was constructed by representing the Tx, Rx, and scatterers as graph nodes. The spatial distance and visibility between nodes were encoded as edge features to describe their geometrical relationships. Based on this representation, an edge-aware graph isomorphism network, termed ScatterGNN, was developed to extract structural features and identify effective scatterers involved in signal propagation. Candidate single- and multi-bounce paths were subsequently generated from the detected scatterers. A path-ranking network, PathRankingNet, was then designed to estimate the validity scores of candidate paths and rank them using a listwise ranking loss. The proposed framework was evaluated in a controlled indoor Industrial Internet of Things scenario generated using Wireless InSite. The scenario covered an area of 150 m × 63 m × 22 m and contained 3,381 Rx sampling locations. Results and Discussions For effective scatterer detection, the proposed method achieved an average precision of 0.9579, while 77.1% of the test samples obtained complete detection of all effective scatterers. For candidate path prediction, the model converged stably after approximately 30 training epochs. Precision@3 ranged from 0.60 to 0.64, Recall@3 ranged from 0.42 to 0.45, and Hit@3 ranged from 0.90 to 0.92. These results indicate that the environmental graph preserves useful structural information related to propagation-path topology. In particular, the model was able to retain at least one reference propagation path among the three highest-ranked candidates for more than 90% of the test samples. Conclusions The proposed framework provides a graph-based approach for transforming environmental geometry into structured representations of effective scatterers and candidate propagation paths. Rather than replacing ray tracing or channel measurements, the method is intended to reduce the candidate search space before detailed path-parameter calculation or channel reconstruction. The current results demonstrate its feasibility within a single simulated environment and primarily reflect its ability to approximate the path-topology labels generated by Wireless InSite. Further validation using independent environments, measured channel data, and path-level parameters such as power, delay, and phase is required before its cross-scenario generalization and practical applicability can be established.
- Environmental graph structure /
- Graph Neural Network (GNN) /
- Effective scatterer /
- Propagation path prediction

HTML全文

图 1 无线环境与信道间的映射关系

下载: 全尺寸图片幻灯片

图 2 无线环境中的散射体划分

下载: 全尺寸图片幻灯片

图 3 有效散射体与路径预测模型架构

下载: 全尺寸图片幻灯片

图 4 有效散射体与路径预测模型

下载: 全尺寸图片幻灯片

图 5 仿真场景

下载: 全尺寸图片幻灯片

图 6 Top-6有效散射体检测结果

下载: 全尺寸图片幻灯片

图 7 训练损失与核心评估指标曲线

下载: 全尺寸图片幻灯片

表 1 训练超参数

参数	值
学习率	1×10^–3
训练轮数(Epochs)	30
批次(Batch Size)	1
损失函数(ScatterGNN)	BCE With Logits Loss
损失函数(PathRankingNet)	Listwise Ranking Loss
优化器	Adam
随机种子	42
数据集分割(训练/测试)	80%–20%

下载: 导出CSV

表 2 核心性能指标

评估指标	初始值(Epoch0)	峰值(对应 Epoch)	稳定值(Epoch 20～29)
Precision@3	0.5332	0.6425(Epoch 17)	0.60～0.64
Recall@3	0.3106	0.4489(Epoch 21)	0.42～0.45
Hit@3	0.8301	0.9202(Epoch 21)	0.90～0.92

下载: 导出CSV

表 3 模型复杂度与预测时延

方法	计算环节	参数量	单Rx耗时
所提模型	散射体检测	2.00 M	11.15 ms
所提模型	路径检测	0.08 M	31.83 ms
所提模型	端到端预测	2.09 M	42.98 ms
ray_tracing	路径求解	-	252.88 ms

下载: 导出CSV

参考文献(27)

[1]	张建华, 王珩, 张宇翔, 等. 6G信道新特性与建模研究: 挑战、进展与展望[J]. 中国科学: 信息科学, 2024, 54(5): 1114–1143. doi: 10.1360/ssi-2023-0355. ZHANG Jianhua, WANG Heng, ZHANG Yuxiang, et al. Channel characteristics and modeling research for 6G: Challenges, progress, and prospects[J]. SCIENTIA SINICA Informationis, 2024, 54(5): 1114–1143. doi: 10.1360/ssi-2023-0355.
[2]	WANG Heng, ZHANG Jianhua, NIE Gaofeng, et al. Digital twin channel for 6G: Concepts, architectures and potential applications[J]. IEEE Communications Magazine, 2025, 63(3): 24–30. doi: 10.1109/MCOM.001.2400213.
[3]	ZHANG Jianhua, WANG Heng, ZHANG Yuxiang, et al. Channel characteristics and modeling research for 6G: Challenges, progress, and prospects[J]. SCIENTIA SINICA Informationis, 2024, 54(5): 1114–1143. (查阅网上资料, 本条文献为第1条文献的英文翻译, 请确认) doi: 10.1360/ssi-2023-0355.
[4]	WANG Jialin, ZHANG Jianhua, ZHANG Yuxiang, et al. Radio environment knowledge pool for 6G digital twin channel[J]. IEEE Communications Magazine, 2025, 63(5): 158–164. doi: 10.1109/MCOM.003.2400168.
[5]	WANG Jialin, ZHANG Jianhua, SUN Yutong, et al. Electromagnetic wave property inspired radio environment knowledge construction and artificial intelligence based verification for 6G digital twin channel[J]. Frontiers of Information Technology & Electronic Engineering, 2025, 26(2): 260–277. doi: 10.1631/FITEE.2400464.
[6]	ZHANG Jianhua, YU Li, LIU Shaoyi, et al. Wireless environmental information theory: A new paradigm toward 6G online and proactive environment intelligence communication[J]. Engineering, 2026, 56: 186–200. doi: 10.1016/j.eng.2025.07.028.
[7]	ZHANG Jianhua, CAI Yichen, YU Li, et al. Four steps toward 6G AI-enabled air interface: Wireless environmental information sensing, feature, semantics, and knowledge[J]. IEEE Communications Magazine, 2025, 63(8): 56–62. doi: 10.1109/MCOM.001.2400528.
[8]	ZHANG Na, YU Li, ZHANG Yuxiang, et al. Towards 6G digital twin channel: Wireless environment knowledge online generator as a key enabler[C]. 2025 IEEE/CIC International Conference on Communications in China (ICCC), Shanghai, China, 2025: 1–6. doi: 10.1109/ICCC65529.2025.11149103.
[9]	于力, 张建华, 蔡逸辰. 6G数字孪生信道的三个使能技术: 多模态感知、环境知识和大模型[J]. 中兴通讯技术, 2025, 31(4): 19–28. doi: 10.12142/ZTETJ.202504004. YU Li, ZHANG Jianhua, and CAI Yichen. Three enabling technologies for 6G digital twin channel: Multimodal sensing, environment knowledge, and large model[J]. ZTE Technology Journal, 2025, 31(4): 19–28. doi: 10.12142/ZTETJ.202504004.
[10]	SUN Yutong, ZHANG Jianhua, ZHANG Yuxiang, et al. Environment features-based model for path loss prediction[J]. IEEE Wireless Communications Letters, 2022, 11(9): 2010–2014. doi: 10.1109/LWC.2022.3192516.
[11]	LIU Fan, CUI Yuanhao, MASOUROS C, et al. Integrated sensing and communications: Toward dual-functional wireless networks for 6G and beyond[J]. IEEE Journal on Selected Areas in Communications, 2022, 40(6): 1728–1767. doi: 10.1109/JSAC.2022.3156632.
[12]	NAGAI K, FASORO T, SPENKO M, et al. Evaluating GNSS navigation availability in 3-D mapped urban environments[C]. 2020 IEEE/ION Position, Location and Navigation Symposium (PLANS), Portland, USA, 2020: 639–646. doi: 10.1109/PLANS46316.2020.9109929.
[13]	SEOW C K and TAN S Y. Non-line-of-sight localization in multipath environments[J]. IEEE Transactions on Mobile Computing, 2008, 7(5): 647–660. doi: 10.1109/TMC.2007.70780.
[14]	WANG Zhonghai and ZEKAVAT S A. Omnidirectional mobile NLOS identification and localization via multiple cooperative nodes[J]. IEEE Transactions on Mobile Computing, 2012, 11(12): 2047–2059. doi: 10.1109/tmc.2011.232.
[15]	WEI Xinning, PALLEIT N, and WEBER T. AOD/AOA/TOA-based 3D positioning in NLOS multipath environments[C]. 2011 IEEE 22nd International Symposium on Personal, Indoor and Mobile Radio Communications, Toronto, Canada, 2011: 1289–1293. doi: 10.1109/PIMRC.2011.6139709.
[16]	ZHANG V Y and WONG A K S. Combined AOA and TOA NLOS localization with nonlinear programming in severe multipath environments[C]. 2009 IEEE Wireless Communications and Networking Conference, Budapest, Hungary, 2009: 1–6. doi: 10.1109/WCNC.2009.4917631.
[17]	DI RENZO M and SONG Jian. Reflection probability in wireless networks with metasurface-coated environmental objects: An approach based on random spatial processes[J]. EURASIP Journal on Wireless Communications and Networking, 2019, 2019(1): 99. doi: 10.1186/s13638-019-1403-7.
[18]	XU Wenkang, XIAO Yongbo, LIU An, et al. Joint scattering environment sensing and channel estimation based on non-stationary Markov random field[J]. IEEE Transactions on Wireless Communications, 2024, 23(5): 3903–3917. doi: 10.1109/twc.2023.3312451.
[19]	XIAO Zhuoran, ZHANG Zhaoyang, HUANG Chongwen, et al. C-GRBFnet: A physics-inspired generative deep neural network for channel representation and prediction[J]. IEEE Journal on Selected Areas in Communications, 2022, 40(8): 2282–2299. doi: 10.1109/JSAC.2022.3180800.
[20]	HUANG Ziwei, BAI Lu, HAN Zengrui, et al. Scatterer recognition for multi-modal intelligent vehicular channel modeling via synesthesia of machines[J]. IEEE Wireless Communications Letters, 2025, 14(7): 1899–1903. doi: 10.1109/LWC.2025.3558059.
[21]	KIPF T N and WELLING M. Semi-supervised classification with graph convolutional networks[C]. 5th International Conference on Learning Representations, Toulon, France, 2017.
[22]	SUN Yutong, ZHANG Jianhua, ZHANG Yuxiang, et al. Environment information-based channel prediction method assisted by graph neural network[J]. China Communications, 2022, 19(11): 1–15. doi: 10.23919/jcc.2022.11.001.
[23]	LIU Lizhou, CHEN Xiaohui, TANG Zihan, et al. PINN and GNN-based RF map construction for wireless communication systems[C]. 2025 International Conference on Future Communications and Networks (FCN), Belgrade, Serbia, 2025: 1–6. doi: 10.1109/FCN66513.2025.11296780.
[24]	GENG Yi, SHRESTHA D, YAJNANARAYANA V, et al. Joint scatterer localization and material identification using radio access technology[J]. EURASIP Journal on Wireless Communications and Networking, 2022, 2022(1): 87. doi: 10.1186/s13638-022-02167-7.
[25]	CAO Zhe, QIN Tao, LIU Tieyan, et al. Learning to rank: From pairwise approach to listwise approach[C]. Proceedings of the 24th International Conference on Machine Learning, Corvalis, USA, 2007: 129–136. doi: 10.1145/1273496.127351.
[26]	张兆灵, 于力, 张宇翔, 等. 无线环境知识表示方法研究——工业互联网场景的应用[J]. 移动通信, 2025, 49(5): 63–66,103. doi: 10.3969/j.issn.1006-1010.20250324-0003. ZHANG Zhaoling, YU Li, ZHANG Yuxiang, et al. Research on representation methods for wireless environment knowledge: Application in IIoT scenarios[J]. Mobile Communications, 2025, 49(5): 63–66,103. doi: 10.3969/j.issn.1006-1010.20250324-0003.
[27]	Remcom Inc. Wireless InSite[CP/OL]. https://www.remcom.com/wireless-insite, 2023. (查阅网上资料,未找到本条文献出版年信息,请确认).