An Improved Three-dimensional Indoor Positioning Algorithm Based on Pedestrian Dead Reckoning

Due to the complex indoor environment of the buildings, Two-dimensional (2D) indoor positioning is difficult to meet the needs of people. Therefore, it is necessary to make some improvements in three-dimensional (3D) indoor positioning. In this paper, a 3D indoor positioning algorithm based on improved pedestrian dead-reckoning (PDR) is proposed. The PDR was improved in motion pattern recognition and step detection. For motion pattern recognition, an improved support vector machine (SVM) method is proposed, which use the confidence threshold to improve the recognition accuracy of the SVM in continuous motion state. For step detection, an improved step detection method is proposed, which sets the peak threshold and valley threshold for acceleration, and the idea of zero-crossing detection method is also used to improve the accuracy of peak detection method. The experimental result of 3D environment shows that the accuracy of motion pattern recognition and step detection is 96.43% and 99.87% respectively and the mean error of the positioning result is 0.63m, which has a reduction of 75.8% compared to the traditional PDR algorithm.