Research On Single Wi-Fi Link Indoor Passive Trajectory Tracking Technology Based On CSI Corrected Phase Information

With the rapid development and wide application of Internet of Things,indoor human tracking plays a very important role in smart home,providing more granular service for the practical application of elderly care,intrusion detection and behavior analysis.However,most of the existing works based on commercial Wi-Fi device require the deployment of multi-pair transceiver devices,and some works also require a large number of offline fingerprint collection,these complex equipment and heavy work hinder the application of indoor human tracking in practice.To address the challenge,we propose our system named Wi-Fi Blinker(WiBlinker),which only requires single pair of commodity Intel 5300 Wi-Fi devices to capture Channel States Information(CSI),by processing and analyzing the phase information of CSI,the moving direction and trajectory of target can be estimated.First,due to the imperfect of hardware equipment,the raw phase information of CSI is polluted by random phase noise,which makes it impossible to apply directly to human tracking.In order to remove it,we propose a novel Dynamic and Light-weight Interpolation Elimination Method(DL-IEM).Second,using the corrected phase information,we propose a Direction Estimation Model(DEM)to obtain the moving direction of target and determine the direction information when the target takes a sharp turn.The key insight of DEM is that the target will not only introduce Doppler frequency shift(DFS)when moving in the monitoring area,but also change the phase difference value between the two adjacent antennas at the receiver.DEM employs DFS to obtain Doppler velocity and coarse-grained direction information for target.Then the Trend of Phase Difference Variation(TPDV)is used to obtain fine-grained direction information.The true direction of moving target can be obtained by combining coarse-grained and fine-grained direction information.Finally,WiBlinker utilizes commercial Intel 5300 NIC to obtain CSI using only one pair of Wi-Fi device without any offline training and without modifying hardware,and the median tracking errors estimated for the straight path,“L” path,and rectangle path are 14.76 cm,21.06 cm and 27.87 cm.