Research On Non-contact Family Health Intelligent Monitoring Algorithm

Non-contact home health monitoring technology has emerged to enable non-invasive monitoring of human health in the daily home environment.Among various non-contact monitoring technologies,Wi Fi technology is superior to others in terms of cost,privacy,accuracy,and popularity.In recent years,researchers have found that channel state information in Wi Fi signals can sense human physiological signals at fine graininess,which might solve the matter of low accuracy of received signal strength.Therefore,it’s of nice sensible price and connexion to review the non-contact home health intelligent observation formula supported CSI.The main work of this paper is as follows:1)In this paper,a CSI-based method for non-contact physiological signal monitoring is proposed.Among the human physiological signals,the respiration and heartbeat signals can best reflect the human health status.This paper firstly preprocesses the collected raw CSI signals and selects the suitable subcarriers by calculating the variance based on the principle that different subcarriers have different sensitivities to physiological signals and the larger the variance is,the easier it is to obtain physiological signals.The discrete wavelet transform and the complete ensemble empirical mode decomposition with adaptive noise were used to separate the respiratory and heartbeat signals respectively,and the respiratory frequency was obtained in the frequency domain by using the direct method.Setting different experimental conditions,the average error of respiratory frequency estimation was finally obtained as only 3.34%.2)Most of the features of heartbeat signals are contained in the QRS heartbeat,and this paper proposes a heart rate abnormality detection algorithm based on dual-slope QRS heartbeat localization.Based on the singularity of the QRS wave,the difference in slope between its left and right sides is used for localization,and a double threshold is set to follow the signal change in real time to ensure the robustness of localization.The MIT-BIH Arrhythmia database was used to verify the effectiveness of the localization and classification algorithm and the sensitivity of QRS wave localization reached 99.65% and the positive prediction rate reached 99.41%.Abnormal classification of heartbeat features by XGBoost algorithm achieves 99.12% accuracy.3)Body movement during sleep is crucial for health monitoring,and it can be combined with other indicators such as respiration and heart rate to monitor sleep and health status.In this paper,the acquired CSI signals are segmented by sliding windows,and the contextual information of the segmented CSI data is learned employing a bi-directional long and short-term memory network(Bi LSTM).The whale optimization algorithm is used to optimize the neural network and extract deeper options to capture the temporal correlation of CSI data.Finally,the softmax perform is employed to spot and classify human motion varieties with a median accuracy of 91%.