Research On Sleep Quality Detection Based On Edge Computing

Sleep is one of the essential activities in people’s daily life.About onethird of one’s life is spent in sleep.High quality sleep can protect mental and physical functions,improve immunity and reduce the risk of disease.Unrestrained and undisturbed detection is the mainstream direction of the development of sleep quality detection technology.Using micro piezoelectric sleep detection mattress for sleep staging has the advantages of non binding,low power consumption and easy access.It is a very good sleep quality detection method to assist hospital examination.However,most of the current sleep quality detection devices are based on cloud computing,so the real-time and consistency of data can not be guaranteed,and the detection accuracy of physiological signals is insufficient,resulting in the lack of accuracy of sleep staging.To solve the above problems,this paper designs and implements a sleep quality detection system based on edge computing.The system adopts the system architecture of "end edge cloud",uses 16 channel piezoelectric ceramic sensors to collect pressure signals,carries out signal preprocessing and rough extraction of data features in the embedded terminal,then sends the extracted parameters to the edge server,discriminates sleep stages in the edge server,and finally realizes data storage through the background system based on Django architecture,The sleep quality test results can be displayed on PC and Android.In the part of physiological signal extraction,aiming at the problems of insufficient accuracy of common extraction methods and serious mode aliasing of empirical mode decomposition,this paper proposes a heart rate extraction method based on variational mode decomposition,which improves the extraction accuracy and solves the problem of mode aliasing.In the part of sleep staging,considering the timing characteristics of sleep staging,the previous time characteristics of the signal are input into the sleep detection model as auxiliary features,and the timing convolution neural network is used for modeling to improve the accuracy of sleep staging.The sleep quality detection system based on edge computing implemented in this paper can not only alleviate the bottleneck of storage and calculation of embedded terminal,but also realize real-time and high-precision sleep quality detection,and provide technical means for sleep disorder screening and sleep health management.