Research On Key Technologies Of Non-contact Sleep Monitoring Based On Doppler Radar

With the accelerated pace of modern life,more and more people are troubled by sleep disorders and related diseases.Therefore,sleep monitoring for disease diagnosis and treatment as early as possible has great significance.The traditional sleep staging method mainly depends on EEG,EOG,EMG and ECG,which has complicated-operation and too many electrodes,costs too much to monitor,bring the tester too much psychological pressure and so on.So research non-contact sleep staging method has great significance.There are many advantages of the sleep monitoring based on Doppler radar.For example,it can be operated simply during long time and without contact,which is potential to be used in family sleep monitoringIn this paper,non-contact sleep monitoring based on Doppler radar is studied.The main work is illustrated as follows:1.The formation and characteristics of EEG signal,the sleep staging criteria based EEG and staging method are introduced2.The basic principle of Doppler radar and non-contact sleep monitoring system are studied,which is based on the physiological signal characteristics to design the filter,and extract respiration heartbeat signal from the radar echo,body movement extracted during resting state and breathing and heartbeat is extracted by valley detection method,simulation of the sleep state,then non-contact sleep monitoring system is verified to be stable during long time and respiration,heartbeat and body movement tested by this system are also verified to be accurate.3.A based on decision tree sleep staging algorithm is proposed,which the establishment of sleep stage decision tree by corresponding relationship between different sleep stages and characteristic parameters which extracted from respiration and heartbeat signal.Sleep staging is completed after selecte the value range of the characteristic.Experimental data show that it can achieve an average accuracy of 59.46%by the proposed algorithm.4.A based on support vector machine sleep stage algorithm is proposed,whichll feature parameters of different sleep stages were extracted from respiration,heartbeat signals and body movement,divide input signal into containing or not containing the heartbeat feature parameters,optimization parameter selection to construct SVM model.Experimental data show that it can achieve an average accuracy of 72.84%by the proposed algorithm.