Research On CSI-based Intrusion Detection And Indoor Localization Technology

In recent years,along with the maturity of wireless communication technology and the wide application Io T technology,people's lifestyle has also changed to intelligent direction.Meanwhile,he demand for intelligent sensing in indoor environments is increasing every day.With the wide deployment of WIFI signals in the world,intrusion detection and indoor positioning based on WIFI have been widely used.It can get good result without professional hardware equipment.Because of this advantage,it has a good application prospect.Different from the traditional WIFI signal strength feature RSSI,this paper selects the finer-grained physical layer channel state information.Based on this information,this article studies the passive intrusion detection and indoor location tracking technology.There are following problems facing: limited feature information used for detection,more reliable detection and judgment method for high dynamic indoor environment,indoor propagation channel seriously affected by multipath effect,and the unreliable accuracy of the single positioning method.In view of the above problems,this paper makes a thorough study of intrusion detection and indoor location technology based on CSI.Firstly,the research background and application significance of environmental awareness are introduced.Then,the current research status of intrusion detection and indoor location technologies is summarized.The problems faced are further studied and improved.Secondly,this paper analyses the algorithms of intrusion detection by three modules: CSI preprocessing,eigenvalue extraction and detection decision according to the execution order.The preprocessing stage focus on the problem that the CSI raw data contains abnormal values and random phase offset due to various noise interferences.For the problem that the amount of CSI data extracted in the existing scheme is limited,this paper proposes an improvement on the existing feature value extraction method.Selecting CSI information from multi-link through correlation analysis.In addition to using the easily acquired CSI amplitude information,simultaneously applying the CSI phase information,then weighted the correlation coefficient matrix eigenvalues of them according to the sensitivity to environmental changes.In this way,the information that used for detection becomes richer and the detection performance can be improved.A density-based clustering algorithm is adopted in the detection decision stage,and an adaptive improvement method is proposed for the parameter setting of DBSCAN clustering algorithm.Finally,a more reliable joint decision result is given by the multi-antenna the receiver.The actual test shows that the detection accuracy of the intrusion detection algorithm in this paper is higher than that of other existing algorithms.Finally,a passive location and tracking scheme based on CSI is proposed.Specifically,for the difficult path identification problem in passive indoor positioning.An identify method is proposed in this paper.The static path component can be removed by conjugate multiplication between antennas firstly.Then the special path can be recognized from the rest paths based on the analysis of the distribution characteristics of positioning parameters by a weight-based reflection path recognition algorithm proposed.After that,accurate positioning parameter can be provided for the next location calculation including AOA,TOF,and DFS.For the problem that the limited positioning accuracy and high requirements for node deployment when using single positioning method,a fusion positioning algorithm is proposed.By combining the estimation results of multi-AP joint AOA positioning with the rough positioning results of the single base station location algorithm,the cumulative error of the rough estimation results can be eliminated.Finally,when tracking the target,the UKF algorithm is used to solve the problem of poor stability of the above positioning results and get closer to the real trajectory.The experiments in actual environment verify the performance of the above-mentioned location tracking scheme.