The Study Of Methods About Indoor Multilayer Localization Based On Multi-sensor Sensing For Smart Devices

With the development of mobile internet and popularity of smart mobile devices, people’s demand for location-based services is increasing, which makes indoor positioning technology become one of the hot field of mobile computing. Since the outdoor positioning which primarily relies on satellite positioning technology is poor in the environment due to satellite navigation signals are rarely accessible blocked by obstacles such as walls, researchers consider using smart mobile devices built-in sensor to improve the robustness of indoor positioning systems and precision. Pedestrian Dead Reckoning is one of the main indoor localization technology, but there is a serious cumulative error and must known initial position. In view of this, we propose a multilayer indoor localization method based on multi-sensor of smart devices which makes full use of pedestrian mobility and dynamic environment to effectively identify and correct user location.Firstly, an improved pedestrian dead reckoning algorithm is proposed through analyzing the gait and direction of change during user walking and using the acceleration cycle change and the direction changes in the corner. The algorithm sets a number of thresholds for detecting the track number of steps and establishes an enhanced stride estimation model by observing large amounts of data. In the moving direction aspect, the algorithm takes turns as anchors and corrects the moving direction through the mean filter. We optimize the parts of pedestrian dead reckoning in order to eliminate the accumulation of errors.Secondly, multilayer indoor localization method based on multi-sensor is proposed through analyzing the characteristics of environment and the user mobility in buildings. The method uses a magnetic field specific indoor environment, with the magnetic field intensity as a position for position matching fingerprints, and uses light features to assist in the detection of the encounter event and then to correct the location and historical trajectory. Using the user mobility to obtain the moving state acquisition sequence, and then combined with the magnetic field location fingerprint matching to determine the floor and initial position so further improve the indoor positioning accuracy and stability while addressing the initial position problem of pedestrian dead reckoning.Finally, different types of mobile phones are used to collect real sensor data and complete experiment is designed to evaluate our method. Then Matlab is used to simulate experiment, which verify the proposed multi-sensor for localization in the indoor multilayer method results.