协方差
人工智能
数学
成对比较
光谱图
模式识别(心理学)
嵌入
算法
鉴定(生物学)
计算机科学
黎曼流形
歧管(流体力学)
相似性(几何)
协方差矩阵
卷积神经网络
推论
无人机
欧几里得空间
标准差
计算机视觉
傅里叶变换
特征提取
特征向量
马氏距离
代表(政治)
人工神经网络
拓扑(电路)
非线性降维
职位(财务)
反褶积
衰减
特征(语言学)
傅里叶分析
作者
Wenze Luan,Liting Sun,Zheng Liu,Xingwei Yan
出处
期刊:Drones
[Multidisciplinary Digital Publishing Institute]
日期:2026-09-18
卷期号:10 (9): 709-709
标识
DOI:10.3390/drones10090709
摘要
Radio frequency (RF)-based drone identification enables passive low-altitude sensing, but its performance degrades under low signal-to-noise ratio (SNR) and long-range propagation. Existing deep models mainly learn spectrogram amplitude textures and underuse the second-order structure and cross-channel correlations of multi-channel RF signals. We propose the Multi-resolution Riemannian-Spherical Network (MRS-Net), a robust identification framework centered on Riemannian Structure Preservation (RSP). RSP derives noise-referenced Riemannian distance, local geometric variation, log-determinant, and multi-scale statistics from local symmetric positive definite (SPD) covariance matrices. Pairwise similarity alignment transfers these structural relationships into the convolutional neural network (CNN) embedding space as a training-stage teacher signal. RSP is removed at inference and therefore adds no online manifold computation. Multi-resolution short-time Fourier transform (STFT) representations provide complementary weak-signal observations, while CosFace enlarges inter-class angular margins. Experiments on DroneRFa and Noisy Drone RF evaluate propagation attenuation and additive noise, respectively. With four-channel DroneRFa input, MRS-Net achieves 90.30% overall balanced accuracy. Compared with SR-CNN-4ch, it improves the 80–150 m result from 61.30% to 78.30%; compared with MR-Spherical-4ch, it reduces the 5-fold standard deviation from 16.78% to 1.46%. Ablations show that RSP is most effective when cross-channel covariance is informative and the backbone preserves local time-frequency structure.
科研通智能强力驱动
Strongly Powered by AbleSci AI