Research On Pedestrian Indoor Fusion Location Method Based On Smart Phone

With the development of mobile internet and the popularity of mobile smart devices,such as smartphones,location-based services become more important.Indoor localization is a key element of location-based services,and current indoor localization technologies cannot meet the needs of practical applications.Using smartphone as the indoor localization device is easy to promote since it does not require additional equipment.Therefore,studying smartphone-based indoor localization is of great scientific significance and applications.At present,single mode-based positioning method has some limitations.Wi-Fi positioning has the disadvantages of low accuracy and poor stability.Pedestrian Dead Reckoning(PDR)has the problem of cumulative error and needs a known starting point.Therefore,this paper proposes an indoor localization method with the smartphones as the platform.The main research contents are as follows:1)In order to solve the problem of Wi-Fi signal intensity uncertainty caused by the difference of equipment,this study proposes Received Signal Strength(RSS)correlation-based Wi-Fi fingerprint localization method.In the offline phase,the Wi-Fi signal intensity information is collected by the smartphones,and the stable Access Point(AP)is selected to construct offline location fingerprint database.In the online phase,the collected APs are filtered based on the APs stored in the fingerprint database.Based on the filtered AP sets,the new offline and online fingerprint database are constructed.Then,the Pearson correlation coefficient and Jaccard coefficient are calculated and used as the similarity parameters.By sorting the similarity parameters,the user's location is determined by K-Nearest Neighbor(KNN)algorithm.Experimental results show that the proposed method can effectively solve the problem of equipment diversity and achieve accurate positioning with an average positioning error of 1.74 meters.2)This paper studies smartphone-based PDR.The peak detection method is used to count the walking steps,and the step frequency model is used to estimate the step length.In order to estimate the heading,we propose a fusion algorithm which can effectively fuse the data of gyroscope and magnetometer.Before the execution of the fusion algorithm,we use the Butterworth low pass filter to filter the data of gyroscope and magnetometer.Then we use the two filtered angle from gyroscope and magnetometer as the input of the fusion algorithm to get the fusion heading.By filtering the fusion heading again,the heading we need is obtained and can be used to detect the pedestrian turning behavior.Experiments show that the cumulative error of PDR is reduced after the above measures,but limited by the sensor accuracy of smartphones and positioning principle of PDR,making the positioning effect is not ideal.Therefore,this paper has carried on the research of the fusion localization.3)This study proposes a method of indoor fusion localization.Based on the proposed Wi-Fi positioning method,the heading and step length information of pedestrian walking,a method is proposed to initialize and obtain the initial position of fusion localization to run the fusion localization algorithm.When the initial position is known,the user's trajectory is reckoned by PDR,and it is corrected by the Wi-Fi localization which consider the heading and the step length of pedestrian walking.Combined with the map information,landmark information and other means,this paper proposes an innovative,effective and high precision indoor fusion localization method.The experiments show that the proposed approach achieve accurate,effective and stable positioning,the average positioning error is about 1.2 meters,and the 80-percentile error of the positioning accuracy is 2 meters.