Research On Indoor Fusion Localization Algorithm Based On WiFi And Inertial Sensor

With the mature of wireless communication technology and the vigorous development of mobile Internet,the demand for location-based Services(LBS)is increasing in daily life.LBS plays an important role in fire rescue,indoor navigation,vehicle navigation and other fields.LBS not only provides convenience for people's life,but also guarantees people's life security.Accurate and efficient access to mobile user location information is the cornerstone of LBS.In recent decades,indoor positioning technology has been developing continuously,among which the most widely used indoor positioning technology is based on WiFi fingerprint positioning.However,the effect of WiFi fingerprint location is not very good under the influence of non-line-of-sight transmission and complex indoor environment in indoor environment.The movement of pedestrians will affect the intensity of WiFi signal received,and a single WiFi fingerprint location system cannot track the trajectory of pedestrians accurately.In order to achieve in the indoor environment of pedestrian movement trajectory tracking accurately,based on the existing positioning technology,this paper proposes a particle filter indoor fusion localization algorithm based on dynamic particle number.This method use the particle filter to combine Pedestrian Dead Reckoning(PDR)and WiFi fingerprint location,and use the modeling information of indoor map to improve the fusion positioning effect.At the same time,the construction method of the fingerprint database in the offline phase of WiFi fingerprint is studied.The specific research contents are as follows:1.The chaotic characteristics of the RSS time series which sampled by the mobile terminal in the indoor environment are analyzed and verified.We use chaotic time series correlation dimension index to determine RSS data quantity which sampled from the AP which replaces the method used in other papers to specify the sample size of RSS data based on experience.In addition,the mapping relationship between the location and RSS is modeled using the gaussian process regression,which reduces the workload of fingerprint collection in the offline phase.2.A step detection algorithm based on adaptive threshold and dynamic estimation algorithm of pedestrian step are proposed to improve the estimation algorithm of pedestrian navigation.By analyzing the acceleration information of pedestrian gait,the speed threshold is dynamically adjusted.The relationship between pedestrian length and stride length and height is analyzed in the two states of normal walking and fast walking.The experimental results show that the accuracy of the proposed algorithm is improved.3.A particle filter indoor fusion localization algorithm based on dynamic particle number is proposed,which integrates PDR and WiFi positioning results to improve pedestrian tracking accuracy.And using the prior information of the indoor map model to solve the problem of track passing the wall.Particle filter particle number is dynamically added and subdivided according to whether the pedestrian is in the turning state,so as to improve the real-time performance of particle filter.The experimental results show that the average positioning error of the fusion positioning system proposed in this paper is within 1 m.The fusion localization algorithm improves the indoor positioning accuracy and enhances the practicability of the positioning system.