| 引用本文： |
-
张萌,王奎,黄伟庆,徐艳云.基于传播场重建与多尺度传播先验融合的室内非合作辐射源定位方法[J].信息安全学报,已采用 [点击复制]
- ZHANG Meng,WANG Kui,HUANG Weiqing,XU Yanyun.Indoor Non-Cooperative Radiation Source Localization Method Based on Propagation Field Reconstruction and Multi-Scale Propagation Priors Fusion[J].Journal of Cyber Security,Accept [点击复制]
|
|
| 摘要: |
| 随着无线通信、物联网和智能终端的广泛部署，室内无线设备及信号活动不断增加，未经授权或异常发射设备带来的安全风险日益突出。面向安全保密监测、异常射频辐射源检测和无线环境感知，准确定位可疑辐射源是风险识别与处置的重要基础。然而，此类辐射源通常缺乏同步、导频等合作信息，同时复杂室内环境中的遮挡与多径效应进一步增加了非合作定位难度。针对复杂室内环境下非合作辐射源定位中传播机理复杂、观测信息受限及模型泛化能力不足等问题，本文提出一种基于传播场重建与多尺度传播先验融合的室内非合作辐射源定位方法。该方法以粗粒度接收信号强度为输入，在场景几何与传播机理先验引导下实现室内传播场的校正与重建，进而提取表征传播机理与空间分布关系的多尺度传播先验，构建传播场表征与辐射源定位联合优化的统一框架。基于实测室内场景的实验结果表明，所提方法能够有效恢复复杂环境中的传播能量分布与空间变化趋势，传播场重建结果的峰值信噪比达到23.2 dB，结构相似性达到92.2%。在定位任务中，所提方法在不同数据划分条件下均表现优越：随机划分条件下平均定位误差和中位数误差分别为1.52 m和1.10 m，分块划分条件下分别为1.87 m和1.50 m。与多种主流方法相比，所提方法在定位精度与鲁棒性方面均表现出明显优势。消融实验进一步表明，所提出的各组成模块均能够对最终定位性能产生积极影响，且完整模型在定位精度上表现最佳。研究结果表明，传播场重建与多尺度传播先验融合能够有效提升复杂室内环境下非合作辐射源定位的精度与稳定性。 |
| 关键词: 室内辐射源定位 射线追踪 接收信号强度 无线传播 深度学习 |
| DOI： |
| 投稿时间：2026-05-11修订日期：2026-08-06 |
| 基金项目: |
|
| Indoor Non-Cooperative Radiation Source Localization Method Based on Propagation Field Reconstruction and Multi-Scale Propagation Priors Fusion |
|
ZHANG Meng, WANG Kui, HUANG Weiqing, XU Yanyun
|
| (Institute of Information Engineering，Chinese Academy of Sciences) |
| Abstract: |
| With the widespread deployment of wireless communications, the Internet of Things, and smart terminals, the number of indoor wireless devices and signal activities has continued to increase, leading to growing security risks from unauthorized or abnormal transmitting devices. For applications such as security monitoring, abnormal radio-frequency source detection, and wireless environment awareness, accurately locating suspicious radiation sources is essential for risk identification and response. However, such sources usually lack cooperative information, including synchronization and pilot signals, while blockage and multipath effects in complex indoor environments further increase the difficulty of non-cooperative localization. To address the challenges of complex propagation mechanisms, limited observations, and insufficient model generalization in indoor non-cooperative radiation source localization, this paper proposes a method based on propagation field reconstruction and multi-scale propagation prior fusion. The proposed method takes coarse-grained received signal strength observations as input and performs calibration and reconstruction of the indoor propagation field under the guidance of scene geometry and propagation-mechanism priors. It then extracts multi-scale propagation priors that characterize the relationship between propagation mechanisms and spatial distributions, thereby establishing a unified framework for the joint optimization of propagation field representation and radiation source localization. Experimental results obtained in a measured indoor environment show that the proposed method can effectively recover the propagation energy distribution and spatial variation trends in complex environments. The reconstructed propagation field achieves a peak signal-to-noise ratio of 23.2 dB and a structural similarity index of 92.2%. In the localization task, the proposed method demonstrates superior performance under different data partition settings. Under random partitioning, the mean and median localization errors are 1.52 m and 1.10 m, respectively, whereas under block-wise partitioning, they are 1.87 m and 1.50 m, respectively. Compared with several mainstream methods, the proposed method exhibits clear advantages in localization accuracy and robustness. Ablation experiments further demonstrate that each component contributes positively to the final localization performance, and that the complete model achieves the best localization accuracy. The results indicate that propagation field reconstruction and multi-scale propagation prior fusion can effectively improve the accuracy and stability of indoor non-cooperative radiation source localization in complex environments. |
| Key words: indoor radio source localization ray tracing received signal strength wireless propagation deep learning |
