Radio Tomographic Imaging Device-free Localization Technology Research Based On Passive RFID Network

Device-Free Localization(DFL)refers to the positioning method which does not need to carry any auxiliary equipment,also known as passive positioning.Compared with the active localization method,passive localization has great application prospects in many fields,such as intrusion detection,emergency rescue and smart home.Radio Tomographic Imaging(RTI)is a novel passive location algorithm.The basic process of RF tomography algorithm is divided the positioning area into a lot of grid of equal size,and then and then map attenuation of the signal strength of the wireless communication link to the attenuation of the grid,and finally showing the localization results in the form of images.Many scholars at home and abroad have done a lot of research and improvement on this algorithm.However,most of these researches are based on Wireless Sensor Network(WSN),and when the localization area becomes larger,the required sensors are more.WSN is probably not the best choice from the cost of building the network and the ease of installation of the sensor.Passive RFID system is the core technology in the Internet of things,composed of readers and tags.Because of its low cost passive tags,the characteristics of small size,in recent years have been widely applied in many fields(such as access control,food traceability,library,etc.)and the communication between the reader and the tag is composed of wireless signal as the medium,so this paper chooses the RFID passive network as the research foundation.However,in the passive RFID scene,the RTI algorithm has a sharp decline in the accuracy of the multi-target location problem,mainly because there are redundant pseudo target points in the imaging results.This paper puts forward two solutions to this problem.The first scheme is to directly extract the local maximum region in the imaging results,and then the data were trained using the Adaboost classifier in machine learning,finally realizes the identification of true target and false target;the second scheme is propose a multi target recognition method based on Cross-Sectional Scanning(CSS).The method firstly analyze and summarize the gray distribution features of local maximum region on imaging results,then structure the Naive Bayesian classifier to eliminate false target,finally obtained the position and number of the real target.At the same time,the method can also recognize the numberof targets when the distance between two targets is very close.In the indoor environment of the experimental results show that the first scheme can effectively identify three targets,and the the target recognition accuracy reached 86% when the positioning errors is in 0.7 meters.Scheme two at least effectively identifies four targets,with an average recognition accuracy of 80%.