Research On Detection And Application Of Personnel Activity Based On WiFi

With the popularity of wireless devices and the opening of wireless network card physical layer information,the new application of WiFi signals has received continuous attention from industry and academia.Personnel activity detection plays an important role in smart home,elderly monitoring and intelligent security.WiFi-based human activity detection method is not affected by light and occlusion,which can make up for the technical deficiencies of existing human activity detection methods.Therefore,WiFi-based personnel activity detection has important research value.The maim contributions of this paper are as follows:1.In the laboratory environment,the channel state information of WiFi signals is collected by self-made spectrum setnsor nodes.This paper observes the correlation between the channel state information of WiFi signal and human activities through experiments.2.Channel state information includes signal amplitude and phase information of antenna subcarriers,but it is difficult to extract phase information.In this paper,the amplitude and phase information are extracted from the original channel state information by data preprocessing.Based on the correlation between channel state information and human activities observed in experiments,two phase preprocessing methods are compared.In the subsequent experiments,a more suitable phase extraction method is selected according to different human activities.3.Combining channel state information feature extraction with various machine learning recognition methods,we have completed intrusion detection experiments,human fall detection experiments,laboratory population statistics experiments and equipment identification experiments.In the intrusion detection experiment,98.33%of the intrusion detection accuracy is achieved.In the intrusion detection experiment,98.33%of the intrusion behavior recognition accuracy was achieved.In the human fall detection experiment,97%of the fall recognition accuracy was achieved.96%of the population recognition accuracy was achieved in the laboratory population statistics experiment.In the equipment identification experiment,a recognition rate of 98.8%was achieved.4.On the basis of completing the offline experiment,the development of the WiFi-based personnel activity detection system was carried out,and the online intrusion detection function was realized.