Research On Target State Recognition Method Based On Wireless Sensing Model

In recent years,with the development of wireless technology,various wireless devices have been deployed in people's living spaces.Wireless sensing technology uses the ubiquitous wireless signals to recognize the status of the target,e.g.,identity,gesture and activity,by analyzing the impact of the target on wireless signals.Thanks to large number of wireless devices,the deployment cost of wireless sensing technology is very low,and it has the advantages of immune to light,high security and no invasion of human privacy so that it can greatly improve the effective of social production.With the development of smart home and human computer interaction,wireless sensing technology has developed rapidly in recent.Nevertheless,there are still many problems in wireless sensing technology.First of all,a large number of wireless sensing works are based on machine learning.However,the wireless sensing works are not robust enough because wireless devices have many kinds of deployment areas and wireless signals are easily affected by the environment.Compare with machine learning based wireless sensing method,model based wireless sensing method has many advantages.On one hand,it is robust for environment,on the other hand,it also helps researchers understand the principles of wireless sensing.This paper proposes a dynamic differential phase model to analyze the dynamic signal caused by the target,and uses the dynamic differential phase as an observation for sensing tasks.For the different areas in the space and different deployment of devices,we propose a dynamic differential phase change rate model,and regard it as a metric to measure the performance of the wireless sensing system.Using this metric,we can select the best sensing area when the wireless devices have been deployed.When the devices have not been deployed,we can improve the sensing performance through enlarging the distance between two adjacent antennas in array.The model we proposed can not only improve the sensing performance of wireless sensing system,but also has a guiding role in the deployment of wireless devices and selection of sensing areas,and would help researchers better understand the principles of wireless sensing technology.Secondly,because of the large coverage of wireless devices,wireless sensing system usually faces with the problem of multi-person scenario in the real scenario.Based on the propagation characteristics of the wireless signal,the signal received from receiver is the composite of multiple reflection signals caused by multiple targets.This would make wireless sensing system cannot work well.This paper proposes a wireless sensing system based on a multi-domain analysis method.By observing wireless signals in multiple domains,the reflected signals caused by multiple persons would be separated,and then the system can work in a multiperson scenario.This paper designs respiration monitoring experiments and angle of arrival estimation experiments to verify the dynamic differential phase and change rate models.The experiments show that the model can improve the performance of the wireless sensing system,and has a guiding role in the deployment of wireless devices and selection of sensing areas.In addition,through multi-person respiration monitoring experiment,the feasibility of implementing a multi-person wireless sensing system based on the multi-domain analysis method is proved.