Device-free Localization Research And Realization Based On WI-FI

With increasingly complex and sophisticated activities in the indoor area, the demands of localization based services are growing. Some research institutions predict the LBS market will sustain a compound annual growth rate of 30% in the next couple of years. With the in-depth study of indoor localization, the device-free localization outstands because of its usability in different application scenarios. Researchers use dedicated hardware or exhaustive fingerprinting to locate the objects to reach high accuracy. But dedicated hardware costs much and fingerprinting gets low efficiency are the heel of Achilles'.For these shortages, this thesis introduces CWF, a system that locates objects by analyzing the signals'Channel State Information(CSI). CWF is a device-free indoor object localization system that provides sufficient accuracy with the commercial off-the-shelf(COTS) transceivers and without requiring neither the objects carrying any wireless devices nor fingerprinting. This system can be detailed as follows:In signal processing part, we propose a method, to deal with multipath effects, called SRCS. Since we can't distinguish the multipath information in the indoor environment, we need to weaken it. This method concludes the multipath mitigation method SRC and selecting the subcarriers. And we introduce a method to match the CSI data with the RSS model according to the relationship between the CSI and the RSS. To meet the need for different objects, the Exponentially Weighted Moving Average(EWMA) is used to estimate the optimized parameter.In object localization part, we introduce DLWF method, which can detect whether the objects in the location area and whether they are in first Fresnel zone. Then we combine the existing localization methods to locate the objects in different area. And we give the analyses about different scenes and prove the system can be used in some muti-person localization and non-of-sight localization scenes.According to the experiments in different conditions, we find that the CWF can reach the target localization without fingerprinting and increase the efficient. And the accuracy of CWF is better than the ways using RSS and a little lower than the way using fingerprinting on CSI. CWF achieves an average location accuracy of 1.3m.