Data Argumented Indoor Positioning Framework Combining IMU, Wireless Signals, and Map-Aided Pedestrian Dead Reckoning

Indoor positioning has become increasingly accessible with the widespread use of wireless signal modules and inertial measurement units (IMUs) in mobile devices. However, achieving accurate localization remains challenging when relying on a single data source. To address these challenges, we propose a hybrid system that integrates data augmentation in both wireless signal-based fingerprinting and IMU-based Pedestrian Dead Reckoning (PDR). Wireless signals are used for initial positioning, while PDR handles updates during movement, with a map-aided PDR (mPDR) approach correcting cumulative PDR errors. The system also incorporates data augmentation techniques to enhance the training data for both wireless signal and IMU-based methods. By dynamically switching between wireless signals and PDR based on user movement status, the system maintains high accuracy even in dynamic environments. Experimental results show that combining mPDR with BLE wireless signals achieves a Mean Distance Error (MDE) of 0.28 meters, significantly outperforming single-source methods. This approach effectively mitigates challenges such as environmental variability and cumulative errors, delivering a robust and reliable indoor positioning solution.