Enhancing indoor positioning system accuracy using a Feedback- Enhanced Federated Kalman Filter (FE-FKF)

Indoor positioning systems (IPS) suffer from errors due to signal interference, sensor inaccuracies, and environmental changes. The challenges associated with the accuracy of the Indoor Positioning System (IPS) through the Received Signal Strength Indicato (RSSI) are plenty. The raw signals are prone to signal interference and multipath fading, and the quality of the received signal is severely impacted in indoor setups due to the internal structure of the walls and the presence of concrete walls, iron structures, metallic furniture, and different electrical wiring objects. All these factors lead to the deterioration of the quality of the signal received. In this paper, we illustrate a novel approach to position estimation based on a Feedback-Enhanced Federated Kalman Filter (FE-FKF) that enhances the accuracy of the Indoor Positioning System work on Federated Kalman Filter (FKF). The novel approach compares the estimated position with a predefined map and gives feedback to the Master filter to update the position. when an anomaly is identified by the error detection mechanism, feedback is relayed to the filter to tune its future estimations by incorporating the rectification request received. In this way, this approach rectifies the errors dynamically, thereby increasing the overall accuracy of the IPS. We evaluate the results of the proposed FE-FKF approach with other methods, such as the Federated Kalman Filter (FKF) on “Position-Annotated-BLE-RSSI-Dataset” from Kaggle. The comparative study of the results concludes that the Mean-Absolute Error (MAE) is significantly reduced with the FE-FKF approach by 18%. Additionally, the inclusion of floor map constraints during the Error Detection Mechanism shows that the position estimation resulting from the proposed study is closely associated with real-world layouts, with improved system robustness and reliability, thereby enhancing applicability in different indoor setups.