Research On Indoor Positioning Technology Of Wi-Fi Integrated PDR

Positioning information plays a very important role in public safety,production safety,the establishment of smart city and some special fields.As the most mature and widely used positioning system,global positioning system(GPS)can provide users with accurate location information outdoors.Indoor due to the shielding of walls,windows and other buildings,the satellite signal is greatly weakened and cannot be located effectively.For indoor positioning technology,a large number of research results have emerged,such as bluetooth,ultra-wideband,geomagnetism,wireless fidelity(Wi-Fi),Zig Bee and pedestrian dead reckoning(PDR).However,most rely on a single signal source,limited positioning area,serious loss and low accuracy.The location technology based on wireless signal fluctuates greatly due to the influence of multipath effect and non-line-of-sight propagation;The indoor positioning technology based on inertial measurement unit cannot get the absolute position information of users.Therefore,the indoor positioning technology with universality,stability and high precision is in the research stage.With the popularity of smart terminals,smart phones with multiple sensor modules have become portable experimental platforms.Taking the built-in sensor of smart phone as the carrier and Android 10.0 smart phone vivo s7 as the platform,this paper studies the indoor positioning technology integrating Wi-Fi and PDR,and studies the Wi-Fi fingerprint positioning algorithm,PDR algorithm and fusion algorithm respectively.The main contents are as follows:(1)This paper discusses the principle and method of Wi-Fi fingerprint location,and explores the fluctuation characteristics of Wi-Fi signal by arranging access points(AP)in different environments and automatically accessing the existing Wi-Fi signal in the environment.In the off-line data preprocessing stage,the average substitution method is used to complete the fingerprint database,the samples are divided into subsets with different characteristics according to the method of obtaining the maximum information entropy,and the random forest(RF)model is used for training,so as to improve the inclusiveness of the samples.In the online matching stage,the real-time signal is matched with the trained samples to obtain high positioning accuracy.(2)The principle of each part of PDR positioning is deeply analyzed.For the step frequency detection part,the peak detection method is optimized,and the method of acceleration threshold constraint and time threshold constraint is used to eliminate pseudo peaks;Step estimation uses the correlation between pedestrian acceleration and pace to fit the wienberg nonlinear step model for estimation;The heading estimation part complementarily integrates accelerometer,electronic compass and gyroscope to obtain the heading angle of pedestrians.Through the experiment,the relatively accurate walking track of travelers can be obtained.(3)In order to realize the complementary advantages of a single positioning method,extended Kalman filter(EKF)is selected to fuse Wi Fi and PDR algorithm.Through the experimental fusion of prediction and observation results,the two positioning results are corrected to each other and more accurate position information is obtained.After comparison,the fusion algorithm can overcome the limitations of a single location algorithm,and improve the location accuracy to 1.26 m.The fused trajectory is basically consistent with the real trajectory of pedestrians.