Study On The Method Of Extracting Vital Signs Based On CSI

In order to predict human health status and guarantee life safety in daily life,noncontact vital sign monitoring technology has been developed.WiFi has the characteristics of wide coverage,simple deployment and low cost,and can realize the extraction of vital signs without contacting the human body.RSS is first used for perception in WiFi,but it has the disadvantages of coarse granularity and low detection accuracy.In recent years,researchers have discovered that CSI in WiFi can be perceived at a finer granularity,and can solve the problem of low accuracy of RSS information detection.Therefore,the study of CSI-based vital sign extraction method has great practical value and important significance.The main work of this article is as follows:(1)A method of extracting vital signs based on CSI signal is proposed.Among the physiological parameters of the human body,the heart-lung signal is very representative.In this thesis,the difference between RSS and CSI information is analyzed.Combined with biomedical technology,a CSI based algorithm for extracting respiratory rate and heart rate is proposed.Use WiFi device and Intel 5300 network card to communicate with the driven notebook,collect CSI data packets,and preprocess CSI signals to eliminate most of the noise interference.Because the mixing of respiration and heartbeat signals is not easy to distinguish,different separation methods and feature extraction algorithms are used according to the waveform characteristics.(2)According to the different sensitivity of different subcarriers to respiratory signal,a more strict triple subcarrier selection algorithm is proposed in this thesis.The three optimal subcarriers are selected to extract the respiratory frequency,and then averaged to obtain the final respiration frequency.Compared with directly using the single best subcarrier extraction,the accuracy of this extraction algorithm is higher.The waveform of the heartbeat signal is relatively complex.The key to extraction is to identify the QRS wave.In this thesis,a QRS wave recognition method based on improved wavelet transform is proposed.After the R peak position is determined,the heartbeat frequency can be directly calculated according to the waveform characteristics of the heartbeat signal.The effectiveness of the recognition algorithm is proved by the MIT-BIH Arrhythmia database.(3)Based on the existing experimental conditions,the actual monitoring platform is developed to verify the feasibility of the algorithm.Through the laboratory and other actual scene tests,the optimal extraction accuracy of the algorithm is respiratory rate 94.6% and heart rate 90.2%,which realizes the monitoring of respiratory rate and heart rate.Compared with other extraction algorithms,the accuracy and robustness of this algorithm are proved.There are 43 figures,8 tables and 82 references in this thesis.