Research And Implementation Of A New Sleep Monitoring System Based On Channel State Information

Due to increasing stress in modern life,more and more people cannot get good sleep quality and suffer from sleep disorder.In-home sleep monitoring systems can help to evaluate sleeping quality and diagnose possible health problems for individuals,which has attracted numerous research interests.Traditional in-home sleep monitoring systems face many problems,such as equipment cost,privacy invasion and sleeping quality deprivation.Nowadays,CSI-based home sleep monitoring systems have drawn considerable attention because of the large-scale deployments of Wi Fi network.However,the breathing rate estimation accuracy and sleeping posture recognition accuracy of existing CSI sleep monitoring system is not satisfying.To solve this problem,this thesis designs and implements a new sleep monitoring system based on channel state information of domestic Wi-Fi network to monitor both breathing rate and sleeping postures of sleepers.The main work of this thesis is as follows:1)In order to accurately estimate the breathing rate of sleepers,this thesis designs and implements a breathing rate estimation subsystem based on new subcarrier selection and peak identification method.After filtering the outliers of the original CSI amplitude sequence with Hampel Filter,a sliding window is utilized to extract the CSI sequence corresponding to the static postures of the sleeper,and a band-pass filter is applied to filter out the noise.Then this thesis proposes a new subcarrier selection method to select the best subcarrier to get the CSI amplitude sequence.Finally,a new peak identification method is proposed to estimate the breathing rate of the sleeper.2)In order to accurately recognize the sleeping postures of sleepers,this thesis designs and implements a new sleep posture recognition subsystem based on PSD peak power ratio.The sleeping postures to be recognized include supine,prone,and side lying.Side lying posture can be directly recognized by the difference of CSI amplitude range.Since supine and prone postures have similar impact on the wireless channel,the recognition between supine and prone is a big challenge in our subsystem.To solve this problem,this thesis analyzes the CSI amplitude sinusoidal distortion differences of supine and prone and then innovatively proposes CSI PSD Breathing-band Peak Power Ratio(PBPPR),which can be utilized to accurately recognize supine and prone sleeping positions.Based on PBPPR,this thesis proposes CSI PSD Breathing-band Peak Neighboring-zone Power Ratio(PBNZPR)to improve the recognition accuracy.Based on the maximization of antenna power ratio,a new antenna selection method is proposed to make the most of the antenna space diversity and improve the recognition accuracy of supine and prone sleep postures.3)Compared with other existing systems,our proposed system could increase the breathing rate estimation accuracy of three sleeping postures by 21%,78% and 80%respectively.Meanwhile,the posture recognition accuracy of three postures could be increased by 27%,16% and 22% respectively.Even when the packet transmission rate or the distance between transmitting and receiving antenna is different,the breathing rate estimation error is less than 0.75 bpm and the sleeping posture recognition accuracy can reach as high as81%.The experiment results show that our proposed home sleep monitoring system could accurately estimate the breathing rate and recognize the sleeping postures of sleepers.