Research On Non-contact Wireless Perception Enhancement Technology

The contactless wireless sensing technology based on commercial devices has become one of the popular technologies for intelligent sensing because of its advantages such as cheap hardware,wide deployment and no need for target carrying any device.The research and popularization of contactless sensing technology has created great potential for indoor navigation,smart home,safety monitoring,health monitoring,behavior monitoring and so on.A wide range of wireless signals have been used for contactless sensing,including acoustic signals and various RF signals such as Wi-Fi and RFID.Among them,RF signals have attracted wide attention in real-world sensing applications,because RF signals do not require the target to be within the line of sight of the transceiver compared with traditional camera-based sensing technologies,and RF signals have stronger penetration capability than acoustic signals due to its longer wavelength.In recent years,domestic and foreign scholars have realized a lot of intelligent sensing systems using commodity radio frequency signals.Typically,like indoor localization,daily activity identification,fall monitoring,etc.Meanwhile,there are also many technologies for sensing more challenging subtle movements,such as air handwriting recognition,lip language recognition,keystroke detection,respiration and heartbeat detection.Although many efforts have been made in the field of RF signal based contactless sensing,applying the existing technologies to practical applications still faces significant challenges.First,when we use the cheap commodity devices for sensing,it is difficult to separate the reflected signal of the target,so contactless sensing is generally based on the change of the resultant signal collected at the receiver to analyze the target's behavior.Due to the fact that the resultant signal of multiple paths does not increase or decrease monotonically with the length change of the reflected path,but occurs fluctuating change with the length change of the reflected path,the sensing performance is highly dependent on the location of the target,thus there exits large amount blind spots for sensing.Second,since the contactless sensing depends on analyzing the change of resultant signal that includes target reflection signals,in general,the target reflection signal is much weaker than direct path signal,so that the contactless sensing range is far less than the communication range,making the sensing range very limited.Furthermore,the sensing accuracy is limited under the multipath interference in complex scenes.Third,in order to ensure the low cost and miniaturization of the equipment,cheap and small Io T devices widely used in wireless sensing generally adopt only a single and linearly polarized antenna design,which results in severe physical limit on the sensing quality of the device,namely polarization loss.As a result,the sensing range is extremely sensitive to the direction of the device,and the sensing robustness is poor.This paper integrates signal processing,model construction,machine learning and hardware design technologies to solve the problems of blind spot,limited range,limited accuracy and unstable range of contactless wireless sensing,and enhance the sensing ability of contactless wireless sensing.The main contributions of this paper are summarized as follows:(i)To overcome the blind-spot issue associated with tiny motion sensing like respiration monitoring based on commodity devices,this paper presents a blind-spot removal method based on the fusion of multidimensional information.According to the theoretical analysis of the Fresnel zone and model verification,we find the unique complementarity between the amplitude and phase information for sensing.Thus we combine the amplitude and the phase readings of the resultant signals to substantially reduce the amount of the blind spots.Then we incorporate multiple RFID tags to further remove the remain blind spots within the sensing area by taking advantage of the cheapness and flexibility of RFID tag.Hence we effectively solve the blind-spot problem with low cost.(ii)To extend the limited range of existing sensing technology,this paper proposes a widerange and through-wall contactless target detection technology by combing Lo Ra signal and drone.Different from the existing technologies that use multiple devices or relay to expand the sensing range,this paper aims to realize wide-area sensing with only one transceiver pair.We utilize the Lo Ra signal to achieve a larger range of sensing,and further incorporates drone's mobility to broaden the sensing area by mounting a single pair of Lo Ra transceiver on the drone for flight scanning.By applying the design idea of reconfigurable antenna to wide-area wireless sensing,we can decompose multiple targets in space domain into single targets in time domain,so that we can overcome the pseudo-target interference from uninterested area.Finally,we achieve wide-area and robust contactless sensing with only a single transceiver pair.(iii)To improve the localization accuracy in complex scenarios with severe multipath interference.This paper presents a contactless robust localization model based on power spectral density,and eliminates the serious multipath interference and localization ambiguity in complex scenes by effective estimation of angle correlation information.Unlike the traditional localization techniques based on attenuation model,angle of arrival and time of flight,this paper proposes to use only the sensitive amplitude information for localization under the condition that the available sensing information of commodity device is very limited.We achieve localization by extracting angle related information from amplitude measurements.We deal with the severe localization ambiguities associated with only a single transceiver pair by using the time correlation of human target movement.And we address the serious multipath interference issue according to the propagation attenuation rule of multipath signals,thus ensure the sensing accuracy.(iv)To avoid the polarization loss of cheap commercial devices that limits the sensing quality,this paper introduces a low-cost link optimization approach based on metasurface.In widely used 2.4GHz ISM band,we design a tunable metasurface with the ability of polarization rotation to adjust the polarization state between transmitter and receiver,thereby reducing polarization loss.We optimize the metasurface structure for both low cost and high transmission efficiency according to the microwave attenuation theory.By synchronizing the receiver sampling and metasurface control,we can integrate the metasurface into the end-to-end wireless link and optimize the link in real time,thus improving the communication quality and enhancing the sensing robustness.