WiFi CSI-based Indoor Localization And Tracking

Recently,with a large-scale deployment of 5G mobile communication network,wireless indoor localization becomes a key enabling technology for ubiquitous location based services in a wide range of applications,such as indoor localization and navigation,emergency rescue,smart home,medical healthcare,security surveillance and so on.Since there are many obsta-cles,RF interference and multipath effect in indoor environments,wireless indoor localization practically confronts many difficulties and challenges.Traditional wireless indoor location sys-tem requires dedicated devices to obtain accurate location information,which results in high cost.Most location solutions with environment dependence are not applicable for the changes of indoor scenarios.RSS-based indoor localization cannot meet the demands of location based services with satisfactory accuracy.To address these issues,this thesis proposes a series of novel approaches on WiFi CSI-based indoor localization and tracking,in terms of pervasiveness,low cost,and high accuracy.The main contributions are summarized as follows.Firstly,due to the environment dependence for most DFL,a novel system has been devel-oped to estimate the target location by refining the angle-of-arrival(AOA)of the target reflection path,which can achieve decimeter-level accuracy in indoor scenarios.With the insufficient an-tennas on commercial WiFi devices,it fully utilizes the phase differences across both different antennas and different subcarriers to enlarge and emulate the number of antennas by spatial s-mooth processing,which can obtain joint AOA and TOA estimation with two-dimensional MU-SIC.Moreover,this DFL system analyzes the characteristics of the static paths and the motion paths,eliminates the static paths with a background elimination algorithm,and separates out the target direct refelction path from the motion paths based on maximum likelihood criterion.According to AOAs of the target reflection path from several WiFi receivers as observation in-formation,the target location can be robustly estimated by particle filter,which effectively gain the accuracy improvement.Secondly,due to the rough accuracy of RSS-based DFL in complicated multipath scenarios,a novel approach of device-free localization and tracking(DFT)is proposed based on target velocity estimation.With mutlpath signal superposition analysis on CSI measurements from commercial WiFi devices,the motion-induced phase shifts are refined from extremely noisy CSI phase,which enhance the robustness of DFT.By phased-array signal processing,joint path length change rate(PLCR)and AOA estimation of the target motion path can be obtained with two-dimensional sparse reconstruction through the spatial sparsity of the target path.With the Fresnel geometrical model between PLCR and the target velocity and location,the fine-grained target velocity can be continuously estimated and then the target motion is tracked with high accuracy.Finally,due to conventional 3D localization technologies with dedicated devices at high cost,a novel 3D localization system is designed and implemented with a breakthrough of two-dimensional AOA estimations,which can bring a new mechanism of 3D localization.Using commercial off-the-shelf WiFi infrastructures,a cost-effective hardware scheme is developed to form L-shaped array,which has an obvious advantage that the azimuth and elevation AOA estimations can be separately obtained using the uniform linear subarray with ordinary AOA esti-mation,effectively reducing the computational complexity.In order to improve AOA accuracy and multipath resolution,CSI subspace with the robustness of noise is figured out by singu-lar value decomposition to obtain two-dimensional AOA estimations,which are formulated as subspace-based sparse reconstruction.According to both azimuth and elevation AOA estima-tions of the target direct path from several receivers,the target can be localized in 3D space in spatial triangulation method with the attractive tradeoff between infrastructure cost and location accuracy.To sum up,this thesis has proposed a series of novel approaches to realize wireless indoor localization and tracking in terms of pervasiveness,low cost,and high accuracy.Plenty of exten-sive experiments in different indoor scenarios confirm the effectiveness of all proposed schemes,which can provide ubiquitous location-based services with high quality.