| 引用本文： |
-
黄潇璇,徐锦龙,沈道源,张萌,魏冬,张巧遇,高宗宁.基于极化指纹空间特性的无人机姿态估计[J].信息安全学报,已采用 [点击复制]
- huang xiao xuan,xu jin long,shen dao yuan,zhang meng,wei dong,zhang qiao yu,gao zong ning.UAV Attitude Estimation Based on The Spatial Characteristics of Polarization Fingerprint[J].Journal of Cyber Security,Accept [点击复制]
|
|
| 摘要: |
| 随着5G与物联网技术的飞速发展，无人机正日益成为空天地一体化网络中不可或缺的智能节点。然而，这一深度融合也使其面临日益严峻的安全威胁。特别是当无人机被恶意劫持时，由于攻击者能够操控无人机的信息传输，导致传统依赖惯性传感器数据的无人机姿态估计方法暴露出显著的安全缺陷。本文首次提出一种基于极化指纹空间特性的无人机姿态估计方案（Polarization Fingerprint Spatial Characteristics-based Attitude Estimation Scheme, PolarSC-AE）。PolarSC-AE方案的核心在于深入挖掘极化指纹（PF）对无人机姿态的敏感特性。与现有的无人机姿态估计方法相比，该方案不依赖通信链路，仅需在基站侧部署正交双极化天线，无需增加任何额外硬件，具备良好的可部署性。我们首先构建了无人机姿态角与极化指纹偏转之间的理论映射模型。针对实际环境中普遍存在的噪声干扰与极化指纹偏转不均匀等问题，我们设计了自适应极化指纹降噪模块（APFD）与多尺度特征对齐模块（MFA），显著提升了特征的鲁棒性与分布均匀性。在此基础上，我们进一步构建了融合可分离自注意力机制的AENet网络，实现了从复杂极化特征到三维姿态角的高精度估计。大量实验验证表明，在1°估计精度与20 dB信噪比的条件下，PolarSC-AE方案的姿态估计准确率高达99.88%；即便在信噪比低至6 dB的复杂环境中，该系统仍能保持86.52%的估计准确率。此外，通过在真实无线场景中的测试，我们进一步评估了该方案在不同环境下的鲁棒性，并详细分析了其计算复杂度及抵御不同攻击手段的能力。综合分析表明，PolarSC-AE不仅具有低成本和易部署的优势，还表现出优异的可靠性和安全性。本研究为无人机的安全、可靠运行提供了创新的感知思路与技术途径，为高鲁棒性、低成本的无人机三维姿态估计提供了切实可行的解决方案，对未来无人机安全架构设计具有重要的参考价值。 |
| 关键词: 极化指纹 无人机姿态估计 |
| DOI： |
| 投稿时间：2025-11-19修订日期：2026-01-30 |
| 基金项目:国家重点研发计划 |
|
| UAV Attitude Estimation Based on The Spatial Characteristics of Polarization Fingerprint |
|
huang xiao xuan, xu jin long, shen dao yuan, zhang meng, wei dong, zhang qiao yu, gao zong ning
|
| (Institute of Information Engineering，Chinese Academy of Sciences) |
| Abstract: |
| With the rapid development of 5G and Internet of Things (IoT) technologies, uncrewed aerial vehicles (UAVs) are increasingly becoming indispensable intelligent nodes in integrated air-ground-space networks. However, this deep integration also exposes them to growing security threats, particularly when UAVs are maliciously hijacked. In such scenarios, attackers can manipulate the information transmission of UAVs, revealing significant security flaws in traditional UAV attitude estimation methods that rely on inertial sensor data. This paper proposes, for the first time, a Polarization Fingerprint Spatial Characteristics-based Attitude Estimation Scheme (PolarSC-AE). The core of the PolarSC-AE scheme lies in thoroughly exploring the sensitivity of polarization fingerprints (PF) to the spatial atti-tude of UAVs. Compared to existing UAV attitude estimation methods, this scheme does not rely on communication links, requires only the deployment of orthogonal dual-polarized antennas on the base station, and adds no extra hardware, ensuring excellent deployability. We first establish a theoretical mapping model between UAV attitude angles and PF deflection. To address issues such as noise interference and uneven PF deflection in practical envi-ronments, we design an Adaptive Polarization Fingerprint Denoising (APFD) module and a Multi-scale Feature Alignment (MFA) module, significantly enhancing feature robustness and distribution uniformity. On this basis, this paper further constructs AENet incorporating separable self-attention mechanisms, achieving high-precision esti-mation of 3D attitude angles from complex polarization features. Extensive experimental validation shows that un-der conditions of 1°estimation accuracy and a 20 dB signal-to-noise ratio (SNR), the PolarSC-AE scheme achieves an attitude estimation accuracy of up to 99.88%. Even in complex environments with a SNR as low as 6 dB, the sys-tem maintains an estimation accuracy of 86.52%. Furthermore, through testing in real wireless scenarios, we further evaluate the robustness of this scheme in different environments and provide a detailed analysis of its computa-tional complexity and ability to resist various attack methods. Comprehensive analysis demonstrates that Po-larSC-AE not only offers advantages of low cost and ease of deployment but also exhibits excellent reliability and security. This research provides innovative sensing ideas and technical approaches for the safe and reliable opera-tion of UAVs, offering a practical and feasible solution for high-robustness, low-cost UAV 3D attitude estimation, and holds significant reference value for the future design of UAV security architectures. |
| Key words: polarization fingerprint uncrewed aerial vehicle attitude estimation |
