Research On The Algorithm Of Vital Signs Monitoring Based On FMCW Radar

Vital signs monitoring is an important research topic in the field of biomedical engineering.Long-term monitoring is of great research significance,it can provide information for doctors to formulate treatment plans,and it can also be used in smart homes.According to whether the system has contact with the subject,it can be divided into two categories.Traditional contact devices have many defects,such as causing allergies and affecting comfort.On the basis of studying the relevant principles of FMCW radar,this paper conducts research on the decomposition of vital signs and frequency estimation algorithms.Due to the influence of static clutter and random body motion,there is strong noise in the vital sign signal detected by FMCW radar,which requires noise suppression and signal decomposition.In order to improve the accuracy of FMCW radar in non-contact vital signs monitoring,and to expand the application scenarios,this paper’s research content is mainly divided into the following five parts:1.A target localization algorithm with multi-frame amplitude spectrum in joint time windows is designed for the initial suppression of static clutter in the environment,DC offset correction is transformed to circle fitting,an extended Differentiate And Cross-Multiply(DACM)algorithm is applied to further enhance the signal-to-noise ratio(SNR).Finally,the micro-movement signal of the object of interest is extracted.2.A signal decomposition algorithm based on Adaptive Notch Filter(ANF)and Empirical Wavelet Transform(EWT)is proposed.ANF is introduced to notch respiratory harmonics before signal decomposition.After updating the spectral segmentation boundary,EWT is used for signal decomposition,and signal screening can be achieved without the need for time-frequency analysis.3.To make greater use of the multiplicative relationship between the signal and its harmonics and combine the time-frequency domain characteristics,a frequency estimation algorithm based on the harmonic relationships is designed.Firstly,the normalization of heartbeat signal is performed in time domain to estimate the crude heart rate.Then the standard deviation is used to benefit the SNR,the spectrum of heartbeat and its harmonic are weighted separately to achieve a two-step estimation of the heart rate.4.A cough detection algorithm based on the multidimensional information fusion is designed.Based on the characteristics of short-time and high-frequency cough action,the magnitude,index and phase acceleration of the Range bin of interest in the IF signal spectrum are used as the analysis objects,and the short-time energy is detected by adaptive thresholding to achieve cough detection.5.The experiment is divided into two parts,the vital signs monitoring experiment and the cough detection experiment.The first experiment is conducted on 15 healthy adults to demonstrate the ability of radar to acquire signals in the chest wall and back of the subjects,with an average accuracy of 92.6%for respiratory rate and 95.5%for heart rate.Subsequently,experiment is conducted on 7 patients with chronic obstructive pulmonary disease(COPD).The average accuracy of 93.4%for respiratory rate and 93.7%for heart rate.Finally,the cough detection experiment achieves a precision rate of 90.6%and a recall rate of 96.7%.