Research On Radio Tomographic Imaging Model Based On RSSI

Radio tomography imaging(RTI)is one of the main device-free location methods.RTI uses the received signal strength indicator(RSSI)of sensor nodes to indicate changes due to target occlusion,and the target is located by inversion image.The core of RTI is to describe RSSI changes of target occlusion by an elliptic weight model,and then solve ill-conditioned linear equations to obtain the image of the monitoring area.Therefore,the elliptic weight model is one of the key factors to determine the imaging quality and positioning accuracy,and it is also a research hotspot in recent years.In order to improve the imaging quality and positioning accuracy,the elliptical weight model will be deeply studied in this dissertation,so that the pixel weighting and the coverage area in the elliptical region can be adaptively adjusted.The model is also applied to the actual moving target tracking,besides,one target trajectory fitting and filtering algorithm is presented to improve the tracking accuracy of moving target in RTI system.The main research work of this dissertation includes the following aspects:(1)The influence of different RSSI features on system performance in RTI technology is studied,including RSSI attenuation feature,RSSI moving variance feature and RSSI histogram distribution kernel distance feature.In view of actual measurement environment,the optimal RSSI feature should be chose as a tradeoff among the positioning accuracy,computational complexity and applicable scenario.Besides,the key parameters affecting the quality of RTI image reconstruction are also analyzed.(2)An adaptive elliptical weighting model based on distance attenuation is proposed,here one distance attenuation factor and elliptical range adaptive adjustment parameters are introduced to improve the standard elliptical model.The introduced distance attenuation factor can make the weightings of grid pixels vary with the distance between the pixel and the LOS path,and match the attenuation law of actual signal propagation;the ratio of the distance between grid pixels and two sensor nodes is defined as an adaptive parameter to realize the adaptive adjustment of ellipse range.In addition,the hardware and software platform of RTI experiment is built.The experimental results show that the proposed model in this paper can effectively reduce artifacts and pseudo-position in the reconstructed image of RTI system,compared with standard elliptic model and constant centrifugal elliptic model.In the experiment,in order to reduce background noise and pseudo-position affecting positioning accuracy,one image denoising method based on NL-means algorithm is presented in the process of image reconstruction.(3)Aiming at the problem of high computational complexity and poor real-time performance of traditional RTI moving target tracking algorithm,a new RTI moving target tracking algorithm based on Savoitzky-Golay filter is proposed.The least square method is used to optimize the target trajectory fitting through the moving window.At the same time,one-dimensional Gauss filter is used to denoise the RSSI measurement fluctuation caused by the target motion,so as to further improve the RTI moving target tracking.