吴世明,,
周牧,
谢良波,
王嘉诚
重庆邮电大学通信与信息工程学院 重庆 400065
基金项目:国家自然科学基金(61771083, 61704015)，****和创新团队发展计划基金(IRT1299)，重庆市自然科学基金面上项目(cstc2019jcyj-msxmX0635)，重庆市教委科学技术研究项目(KJQN201800625)
详细信息
作者简介:杨小龙：男，1987年生，讲师，博士，研究方向为无线感知、室内定位
吴世明：男，1994年生，硕士生，研究方向为WiFi穿墙目标检测、人体行为识别
周牧：男，1984年生，教授，博士生导师，研究方向为无线定位技术
谢良波：男，1986年生，副教授，研究方向为虚线射频识别技术
王嘉诚：男，1992年生，博士生，研究方向为室内定位技术、阵列信号处理
通讯作者:吴世明 2812940421@qq.com
中图分类号:TN929.5计量
文章访问数:2046
HTML全文浏览量:776
PDF下载量:82
被引次数:0
出版历程
收稿日期:2019-05-24
修回日期:2019-12-07
网络出版日期:2019-12-14
刊出日期:2020-03-19
Indoor Through-the-wall Passive Human Target Detection Algorithm
Xiaolong YANG,Shiming WU,,
Mu ZHOU,
Liangbo XIE,
Jiacheng WANG
School of Communication and Information Engineering, Chongqing University of Posts and Telecommunications, Chongqing 400065, China
Funds:The National Natural Science Foundation of China (61771083, 61704015), The Program for Changjiang Scholars and Innovative Research Team in University (IRT1299), The General program of Chongqing Natural Science Foundation (cstc2019jcyj-msxmX0635), The Scientific and Technological Research Foundation of Chongqing Municipal Education Commission (KJQN201800625)
摘要
摘要:穿墙场景下，由于墙体造成信号严重衰减，接收信号中目标反射信号的能量大幅下降，接收信号淹没在收发机直射信号和室内家具反射信号中，难以检测墙后目标。针对上述问题，该文提出一种新颖的基于多维信号特征融合的穿墙多人体目标检测算法(TWMD)。先对接收到的信道状态信息(CSI)进行预处理以消除相位误差和幅值噪声，再利用CSI的时序相关性和子载波相关性从相关系数矩阵中提取多维信号特征，最后使用BP神经网络完成特征与检测结果之间的映射。实验结果表明，该算法在玻璃墙、砖墙和混凝土墙环境的识别精度分别在0.98, 0.90, 0.85以上。根据所统计的4000个各类样本的检测结果，与现有基于单一信号特征的检测算法相比，该文算法在对不同数量运动目标的检测上，获得了平均0.45的精度提升。
关键词:无源人体目标检测/
WiFi/
信道状态信息/
多维信号特征
Abstract:In through-the-wall scene, due to the serious attenuation of signal caused by wall, the energy of target reflection signal in the received signal decreases significantly and the received signal is submerged in the direct signal of the transceiver and the reflection signal of indoor furniture, making the target behind wall is hard to be detected. In view of the above problems, a novel Through-the-Wall Multiple human targets Detection (TWMD) algorithm based on multidimensional signal features fusion is proposed. Firstly, the received Channel State Information(CSI) is preprocessed to eliminate the phase error and amplitude noise, and the multidimensional signal features are fully extracted from the correlation coefficient matrix by using time correlation and subcarrier correlation of CSI. Finally, the mapping between features and detection results is established by BP neural network. The experimental results show that the recognition accuracy of this algorithm in the environment with glass wall, brick wall and concrete wall is above 0.98, 0.90, 0.85, respectively. According to the detection results of 4000 samples, compared with the existing detection algorithms based on single signal feature, the proposed algorithm achieves an average accuracy improvement of 0.45 in the detection of different number of moving targets.
Key words:Passive human target detection/
WiFi/
Channel State Information(CSI)/
Multidimensional signal
PDF全文下载地址:
https://jeit.ac.cn/article/exportPdf?id=d2d5e8dd-5e68-44d1-b8c7-e781529a5702
