Human Intrusion Detection And Indoor Localization Based On Channel State Information

With the continuous development of information technology,WLAN is deployed by more and more individuals and organizations in various indoor application scenarios.This has also attracted more and more researchers to use it to carry out some related applied researchs.In many indoor application scenarios,human intrusion detection and indoor localization are extremely valuable for application research.Traditional human intrusion detection and indoor localization mainly use some techniques including RFID,ultrasonic,camera,infrared and so on,but there are also some problems in the accuracy or cost control of the above technologies.However,channel state information(CSI)in WLAN can overcome many problems because of its own characteristics and advantages,so it has gradually become the preferred reference for many researchers.In addition,in order to achieve better performance of human intrusion and indoor localization research,this paper also introduces a visibility graph(VG)method to obtain network features that cannot be extracted by conventional methods to implement intrusion detection and improve localization.Moreover,this kind of viewable networked mapping method brings CSI-based human perception research into a new field,which has contributed to the further development of this research direction.The contributions for this paper mainly are detailed as the follows:(1)Firstly,summarizing and analyzing the traditional human intrusion detection and localization technology,and then pointing out the shortcomings of traditional methods.Finally,analyzing the progress of CSI and complex network;(2)Introducing the specific principle and acquisition method of CSI and the specific principle of the VG method.Moreover,several commonly used machine learning algorithms are briefly introduced;(3)Human intrusion detection is achieved,which uses VG method combined with CSI time series;(4)Extending the VG method from the time series to the subcarrier sequence signal,and constructing a viewable network and obtaining more effective features for improving the indoor localization;(5)Conducting different contrast experiments to explore and analyze the influence of relevant factors on intrusion detection and indoor localization.Moreover,a specific example,by the VG method is shown in order to illustrate this method is helpful for improving the localization effect;(6)Analyzing and comparing the relationship between time and accuracy under different localization algorithms,and giving some suggestions for algorithm selection under different scenarios and requirements.