Design And Implementation Of Indoor Positioning System On Mobile Device

Accurate indoor position for pedestrian plays a more important role in daily life.With the popularity of smartphones and wireless network environments,it has important significance to use smartphones and wireless AP to locate pedestrians,which can be applied to advertising,monitoring and other fields.However,most current indoor location methods still have some problems in convenience and accuracy in actual application scenarios.In order to design and implement indoor location system accurately,two positioning method based on pedestrian dead reckoning and wireless access point are presented in this article.Pedestrian dead reckoning method estimates the number of steps,the length of each step and direction of walking by acquiring data sampled from inertial sensors built in smartphone.The position of each step can be calculated iteratively.The number of steps are counted by peak detection of acceleration data.The length of one step is estimated according to a mathematical model which is trained by step dataset.The direction of each step is calculated by solving the rotation matrix of the phone coordinate system and the world coordinate system.The method based on wireless access point contains online and offline stages.In the offline stage,floor area is gridded and a fingerprint database for wireless access point is designed by collecting AP signals received in each grid.In the online stage,signals of wireless access point received by pedestrians are matched with the constructed fingerprint database,and the location is determined by fingerprint similarity.The indoor floor plan is digitized according to international standards,and the unit of measurement of methods mentioned above is unified using the floor plan,which provides a foundation for the fusion of the positioning results.The developed location system was tested in experimental scenes including a residential building and a school building.The experiment results show that the presented method performs a certain accuracy at short and medium distances.Due to the instability of wireless signals and scene limitations,the robustness of the presented method has not been fully evaluated.In the future,the wireless fingerprint database will be updated by crowdsourcing to improve the location accuracy.