Wi-Fi动作识别在室内入侵检测中的应用

为解决当前入侵检测设备成本高、隐私性差、安全性难以保障的问题,以家用无线路由器为信号源,提出将Wi-Fi动作识别技术应用于室内入侵检测的研究。该系统建立在3D-CNN深度神经网络的基础上,对收集的Wi-Fi信道状态信息进行降噪、滤波、特征提取等一系列处理,将原始信号转化为BVP信号,进一步通过神经网络识别信号对应的入侵行为。实验结果显示,即使在信号干扰较强的室内办公环境中,该系统的入侵检测精度也能够达到100%,入侵行为分类精度不低于98%,表明该系统在室内入侵检测中具有较高的应用价值。

In order to solve the current problems of high cost,poor privacy and difficult security of intrusion detection devices,we propose the study of applying Wi-Fi action recognition technology to indoor intrusion detection using a home wireless router as the signal source.The system is built on the basis of 3D-CNN deep neural network,which performs a series of processing such as noise reduction,filtering and feature extraction on the collected Wi-Fi channel state information,converts the original signal into a BVP signal,and further identifies the intrusion behavior corresponding to the signal through the neural network.The experimental results show that the intrusion detection accuracy of the system can reach 100%even in the indoor office environment with strong signal interference,and the intrusion behavior classification accuracy is no less than 98%,which shows the high application value of the system in indoor intrusion detection.