Study On Indoor Positioning Method Based On Smartphone Multi-sensor Considering Pedestrian Activity

In recent years,people a large of demand for Location Based Services(LBS).Indoor and outdoor positioning technology is particularly important in the application of Location Based Services.In outdoor positioning,it relies on the Global Navigation Satellite System(GNSS)for positioning and Navigation,which has been relatively mature.However,Satellite signals are severely blocked by buildings and are weak when they reach the indoor environment,making it impossible for them to obtain accurate location information in the indoor environment.In indoor positioning,the positioning technology is developing,but the development is not yet mature,more and more people join in the study of indoor positioning technology.With the progress of science and technology and the improvement of people's living standard,smart phones have been popularized,and Wi Fi signals are everywhere.Indoor positioning technology based on smart phones has become a hot research spot.Currently,smart phones can use their own sensors to obtain data,and rely on Pedestrian Dead Reckoning(PDR)technology and Wi Fi fingerprint positioning technology to realize positioning,without the need for other additional positioning devices.Therefore,multi-sensor positioning based on smart phones has become one of the most worthy indoor positioning technologies in the future.However,there are still some problems to be solved in the practical application of these technologies.For example,in a complex indoor environment,Received Signal Strength(RSS)may be affected by body and pose,resulting in signal refraction,reflection and diffraction,resulting in poor fingerprint matching effect and poor positioning accuracy.In the process of fingerprint positioning,how to reduce the influence of body occlusion and mobile phone pose on fingerprint matching;In fingerprint positioning,how to avoid the problem of poor fingerprint location accuracy caused by differences in RSS received by heterogeneous smart phones;How to consider floor information in multi-floor environment and how to accurately identify floors;How to reduce the defect of single PDR and Wi Fi and the influence of pedestrian activity.Based on smart phone carrier,this paper use the Wi Fi,PDR indoor positioning technology and machine learning method,grey correlation theory,the extended kalman filtering theory,through the theoretical research and experimental verification,under the different influence factors is discussed in detail,Wi Fi signals,and the characteristics of the inertial sensor signal,grey relational analysis are studied different terminal from correct fingerprint localization algorithm,several floors environment improving Wi Fi fingerprint localization algorithm of the attitude,and pedestrian activity recognition auxiliary PDR and Wi Fi fusion algorithm.The main contributions of this paper are as follows:(1)analyzing the characteristics of RSS in different positions,body shielding,different pose,heterogeneous smart phones,the existing problems of Wi Fi fingerprint positioning can be found,so as to further improve the accuracy of Wi Fi fingerprint positioning.At the same time,it also analyzes the signal characteristics of the inertial sensors of smart phones under different pedestrian activities,paving the way for the follow-up research on PDR and Wi Fi fusion positioning assisted by pedestrian activity recognition.(2)aiming at the problem of poor positioning accuracy of traditional fingerprint location methods caused by differences in RSS signals received by heterogeneous smart phones,this paper proposed a free correction fingerprint location algorithm based on grey correlation analysis.The algorithm uses grey relational analysis to determine the user location by calculating the grey relational degree between the fingerprint point and the test point.Through multiple sets of experimental data analysis of the proposed method and the commonly used positioning performance of heterogeneous terminal correction methods,the results showed that under the condition of heterogeneous is put forward in this paper the fingerprint orientation method based on grey correlation analysis of average position error,root mean square error and standard deviation are the best,the method has good stability,adaptability and positioning accuracy.(3)This paper proposes an improved Wi Fi fingerprint location algorithm that takes into account the posture in the multi-floor environment.This method takes multi-floor environment into consideration,and proposes a Wi Fi floor recognition algorithm based on Support Vector Machine(SVM).The experimental results show that the recognition accuracy between floors can reach 98.89%.Considering the influence of attitude on the positioning results,a traditional fingerprint positioning algorithm for Wi Fi assisted by attitude recognition was proposed,and the influence of Wi Fi frequency band and column on the positioning results in the environment was analyzed.The experimental results show that the mean error of WKNN,KNN and bayesian algorithm is significantly reduced when attitude recognition is considered.Based on SVM machine learning theory,a SVM fingerprint localization algorithm is proposed in this paper.The results of experiments in multiple environments show that the SVM fingerprint localization algorithm assisted by attitude recognition has high accuracy,good localization performance and adaptability.(4)aiming at the poor positioning accuracy of single Wi Fi and PDR in different activities,this paper proposes a PDR and Wi Fi fusion positioning algorithm for pedestrian activity recognition.Firstly,the common activities in the nine were defined,and the recognition and analysis of the nine activities were carried out through the processes of data preprocessing and feature extraction.The results showed that the average recognition accuracy of the nine activities reached 99.55%.Then the step model of different activities are analyzed.Finally,EKF is used to integrate Wi Fi and PDR technology for pedestrian activity recognition.Through experimental comparison and analysis of the performance of PDR,Wi Fi positioning algorithm and EKF fusion positioning algorithm assisted by pedestrian activity identification,this algorithm reduces the fluctuation of Wi Fi fingerprint positioning results and improves the positioning accuracy of PDR.Therefore,Wi Fi and PDR fusion algorithm have better positioning performance.