An Auto-Adjusting Algorithm to Enhance Indoor Localization Accuracy: A Real-Time Experimental Analysis

Indoor localization systems (ILS) have become essential tools to address the challenge of locating and tracking items and individuals, such as children, the elderly, and patients with Alzheimer’s or dementia. In this study, our aim is to develop an auto-adjusting algorithm to select coefficients based on the current automatically Received Signal Strength Indicator (RSSI) class. The method adopted by the Internet of Things (IoT) model integrated with Bluetooth Low Energy (BLE) to achieve this objective. The system is designed to track lost items and individuals using a wearable central unit (mini-Raspberry Pi) as a controller and BLE nodes as peripheral devices. The developed system includes Bluetooth beacons, data aggregation, storage, and a web interface for real-time tracking and visualization. The RSSI foot printing method is adopted to detect a specific zone within indoor environments. A web-based application has also been developed to enable monitoring and management of the designed system. The study was evaluated in a real-time experimental environment (with fixed and auto-adjust coefficients) to explore the challenges of accurately determining indoor locations in five rooms. The proposed method initially succeeded in reducing the error caused by fixed coefficients and RSSI by 28.03%. The results demonstrated that the auto-tuning algorithm with dynamic coefficients was able to improve the accuracy of positioning by dynamically adjusting RSSI coefficients; this study successfully reduced the average absolute percentage error of indoor localization by 8% and decreased the maximum localization error to 2.01 meters.