The Research And Implementation Of Motion Sensing Method Based On Wireless Data Processing Technology

In recent years,people's consciousness of family security advances with the development of society and technology.Among them,intrusion detection,fall detection and sleep monitoring are three typical applications of family security protection.They will have promising future in smart home,and there are many related products have appeared already.In the field of intrusion detection,most used products are infrared light,magnetic door and video monitoring;In the field of fall detection,most used products are wearable equipment,manual trigger equipment and video monitoring;In the field of sleep monitoring,most used products are PSG,smart sensors,video monitoring and smart phones.However,these products still have some disadvantages in use,such as being easily affected by environmental factors,having the possibility to invade people's privacy or needing to wear some sensors.Channel State Information is the channel state of wireless signal which depicts the multipath propagation to a certain extent.It can characterize the differences between human motions.Also,it has many characteristics,such as anti-jamming,private and non-contacting.Based on these characteristics,this thesis proposes three motion perception algorithms which are used in intrusion detection,fall detection and sleep monitoring.The research content will be illustrated as follows:(1)The thesis proposes a lightweight intrusion detection algorithm which is based on CSI.First of all,the collected data will have a linear transformation to eliminate phase error of CSI;Secondly,two reference points will be defined.One is the short-term effect size to detect whether the man is walking at current.The other is the long-term effect size to detect whether the man is walking continuously;Thirdly,the results of the these two reference points will be combined to judge whether someone is walking in the monitoring area.Experimental results show that the algorithm can detect intrusion quickly.(2)The thesis proposes a fast fall recognition algorithm which is based on CSI.Firstly,the original data will have a data weighting operation to smooth the noise;Secondly,the short-term effect size will be used to extract behaviors data;Thirdly,features of behavior data will be extracted and the KNN classification algorithm will be employed to identify the fall behavior of people.Experimental results show that the algorithm can identify the fall behavior effectively.(3)The thesis proposes a non-contacting sleep monitoring algorithm which is based on CSI.Firstly,the CSI data of sleeping will be collected by three groups of experimental equipment;Secondly,Hampel filter and Butter Worth low-pass filter will be used to eliminate the outliers and noise of the original data;Thirdly,the five nearest subcarriers of thirty subcarriers will be selected to represent the state of breath;Finally,the sleeping postures will be testified by the different degrees of influence which different sleeping postures have on experimental equipment.Experimental results show that the algorithm can monitor breath and sleep postures effectively.Based on these three algorithms above,this thesis has proposed a family security protective system.It can quickly identify and monitor people's invasion,falling and sleep.Furthermore,this system has solved the problems of being easily affected by environmental factors,having the possibility to invade people's privacy and needing some sensors to assist.