Design And Implementation Of Step-signal-based Single-AP Multiple-targets Localization System

High precision indoor localization technology has been widely used in storage,consumption,medical and other fields,such as warehouse goods positioning,exhibits location navigation and patient location monitoring.The traditional indoor localization methods require the target to carry localization devices,positioning the target by getting the location of the devices.In contrast,device-free localization(DFL)methods don’t need extra devices,can be quickly deployed and put into use.However,the current DFL systems have high deployment cost and maintenance cost due to the cooperative work of multiple APs.Among the DFL methods,active DFL needs the active participation of the users,which leads to poor user experience.Although device-free passive-localization(DFPL)does not require active participation of the users,it’s hard to get desired signal because of the low signal-noise ratio(SNR).Therefore,it is challenging to achieve high-precision positioning.In order to meet the challenges,this paper shows a single-AP DFPL & recognition-tracking system based on footstep signal under indoor environment,using different components of the step signal(including structure-borne impact signal and air-borne impact signal)to locate the target,reaches 30 cm median error and can recognize different people.The main work of this paper is as follows:(1)This paper shows the footstep-signal-based single-AP DFPL & recognition-tracking system for the first time.This paper finds the different components of the step signal: airborne impact signal(AIS)and structure-borne impact signal(SIS),and uses multi-channel AIS signals to achieve the estimation of the target’s signal Angle-of-arrival(Ao A)along with recognizing different targets.At the same time,it uses the SIS signal to ranging,reaches30 cm median error and can recognize and track different people at the same time.(2)This paper raises a new spectral weighting algorithm during the Ao A estimation procession,which solves the problem of low Ao A estimation resolution in single-AP positioning systems.Different from the traditional spectral weighting algorithms,the new spectral weighting algorithm realizes adaptive noise processing and performs well under different SNRs.It improves the estimation accuracy of Ao A under single-AP condition,making the estimation error of 80% less than 4°.(3)This paper uses the SIS signal,raises a new ranging method based on adversarial training,which has robustness to environmental changes while ensuring the accuracy of distance estimation.In this paper,distance correlation characteristics of the SIS signals are studied.By using the feature extraction ability and high robustness of the adversarial training network,the system reaches a median error of 30 cm under the single-AP condition and is adaptable to the environment.