WIFI_Relative
level: author: Daqing Zhang Peking University date: ‘2017-09-05’ keyword:
- WiFi activity sensing,Fresnel zone model
Paper: CSI Fresnel Zone
Summary
这篇文章介绍了CSI,Fresnel现象,以及如何通过Fresnel模型 来感知用户微小的动作。
Research Objective
- Application Area:CSI contains detailed phase and amplitude information about each subcarrier and depict each subcarrier’s propagation path,where each subcarrier is affected differently by human activities such as gestures ,falls,respiration.
- Purpose: examine the basic theoretical foundation for RF-based sensing systems,introduce the Fresnel zone model to the indoor environment using it to depict RF signal-propagation properties.
previous work:
- RADAR indoor localization system , multipath cause unpredictable RSS fluctuation.
- path loss model: the reflected path length changed with the person’s motion , leading to RSS fluctuation ,and difficult to quantify each path in indoor multipath rich environments.
- Interference-based propagation model: divides all indoor multipaths into static and dynamic paths,the path length of the dynamic reflected signal changes,the phases of the reflected signals change. the recieved signal’s fluctuation is attributed to the interference of the dynamic signals and the superimposed static signals, quantitative relationship between CSI dynamic and human movement speed is established. it’s still difficult to correlate the CSI waveform to human motions in geometrical space.
Methods
Fresnel zones : the generated Fresnel zones are determined by the aggregated signal vector of all static multipaths, the reflected signals are generally weaker than the LOS signal,making the location and the size of the Fresnel zones slightly different from the standart ones.
【定义问题1】how to characterizing the relationship among multifrequency Fresnel zones?
【定义问题2】how to apply the Fresnel zone to human sensing?
-
Human respiration rate detection :
1.the micro human activity is so weak(<font color=red>about 5mm</font>) that cant be distinguished from environmental noises,user need to sit between the LoS or very close to the transceivers. 2. how to correlating subtle chest displacement to signal fluctuation? the phase rotation is about 60,periodical respiration contain four segment: 1.a wave generated by inhalation,a straight line caused by the pause after inhalation,a waveform generated by exhalation,a waveform generated by straight line caused by the pause after exhalation.![](https://lddpicture.oss-cn-beijing.aliyuncs.com/picture/image-20191108094205335.png) 3. how to leveraging multifrequency Fresnel zones to enhance respiration sensing? <font color=red>这一部分不是很明白?</font> when the subject was located in the middle of one zone,but this signal became increasingly distorted as the subject moved toward the Fresnel zone boundary until finally no clear signal was present.![](https://lddpicture.oss-cn-beijing.aliyuncs.com/picture/image-20191108094350852.png) 2. 【定位问 题2】how to detect walking direction in a 2D plane?![](https://lddpicture.oss-cn-beijing.aliyuncs.com/picture/image-20191108095322842.png) 1. subvector distance estimation? the distance can be expressed as the number of Fresnel zone boundaries crossed,the nunber can be measured by counting the peaks and valleys of the received signal using fast Fourier transform. <font color=red>只通过波峰波谷来测距离 不准</font> 2. subvector direction estimation? using a method to extract this precedence relationship among subcarriers.
Evaluation
- Environment:
Conclusion
-
model can enable centimeter-scale human activity recognition
and potentially expand applications in areas such as healthcare, localization, and human– computer interaction into new territories.
-
Fresnel model shortcoming: if the moving objects are small and far from the transceivers ,the dynamic path signals usually have a negligible effect on the aggregated signal compared to the strong Fresnel zone using major static pahts signals.if all moving objects significantly affect the received signal compared to the LOS signal ,the static Fresnel zone model might not analyze the signal-propagation properties.
Notes
- whether the micro motion is detectable or not depends not only on how strong the reflected signal strength is compared to the direct signal strength but also on the location of human body with respect to the two transcevers.