Research On Indoor Personnel Localization And Recognition Method Based On Millimeter-Wave Radar

With the continuous integration and development of Internet of Things(IOT)and Artificial Intelligence(AI)technologies,IOT has entered a new era of intelligent interconnection of all things.Indoor sensing technology has also transformed from the focus on things connected to the network to the era of human-centered intelligence,and the service concept of wisdom,security and efficiency of indoor devices is deeply rooted.Indoor sensing technology collects and analyzes indoor scenes through various embedded network terminal devices,making various terminals with service "thinking" to provide a more intelligent living environment for human beings.Among many wireless sensing devices,Millimeter-Wave Radar devices have become one of the important tools in indoor sensing with their own advantages.Compared with traditional low-frequency wireless sensing devices,millimeter-wave radar signal wavelengths are in the millimeter range and can perform indoor sensing tasks with finer recognition granularity and higher accuracy.This thesis focuses on localization and identification in indoor environments,and the main research work and contributions are as follows:(1)An indoor localization method based on multipath propagation of millimeter-wave radar signals(MM-IL)is proposed to achieve indoor person and object localization.First,the indoor radar signal propagation path is determined,and the target distances on the line-of-sight and non-line-of-sight paths are extracted by distance-micro-Doppler features;then,a 3D localization model based on the indoor multipath effect is established to realize indoor personnel and object localization,which fully considers the help of multipath effect and uses the reflected signals from the ground and walls to locate the target positions precisely.The problem that the radar in single-input single-output mode cannot obtain the 3D spatial position of the target is solved;finally,it is experimentally demonstrated that the MM-IL method can achieve 3D positioning error within 15 cm within 3.6m.(2)A multi-view gait recognition method based on the millimeter-wave radar(MM-GRM)is proposed.The multi-view gait recognition is achieved using microDoppler gait features on different viewing angles for network training.Firstly,the raw gait data from different walking viewpoints of personnel are collected using millimeter-wave radar;then,the range of personnel movement from different perspectives is estimated by distance-micro-Doppler features,and the distancemicro-Doppler features with the presence of personnel gait information are superimposed to eliminate static background noise and enhance personnel gait features;finally,the gait recognition network model is constructed by the gait data set from multiple viewpoints.The experimental results show that the proposed method achieves an average recognition accuracy of 88.4% indoors and solves the gait recognition problem from multi-views.(3)A millimeter-wave radar-based vital sign recognition method(MM-VSR)is proposed to realize vital sign recognition in long-range LOS paths and NOLS paths.First,the radar collects the raw data of personnel’s vital signs;then,by superimposing four antenna data,the dynamic signal of vital signs in the distancemicro Doppler feature is enhanced,and the static noise in the environment is suppressed,which solves the problem of low accuracy of thoracic localization in the line of sight path and non-line of sight path by the traditional radar respiratory heartbeat monitoring method;finally,using an improved empirical mode decomposition model based on the calculation of respiratory and heartbeat The amplitude contribution rate on the frequency band is used to accurately separate the respiratory and heartbeat data in the vital signs.Experiments have shown that the MM-VSR has a breath monitoring error of fewer than 1.52 breaths per minute within3 meters.