Wearable‐Assisted Localization in Wireless Networks: A Hybrid Approach

ABSTRACT Accurate and real‐time localization in dynamic environments remains a persistent challenge, particularly in scenarios where GPS signals are unreliable or unavailable. To address this limitation, the present study proposes a hybrid wearable‐assisted localization (WAL) framework that synergistically integrates wearable sensors, fuzzy logic, machine learning (ML), and sensor fusion within wireless networks. The novelty of this work lies in its multi‐source adaptive fusion approach, wherein uncertain sensor measurements are first modeled through a fuzzy inference system to mitigate environmental noise and drift. Subsequently, supervised ML algorithms—including Support Vector Machines (SVMs) and Neural Networks—are trained on labeled multimodal datasets combining accelerometer, gyroscope, heart rate, Wi‐Fi, and Bluetooth data to predict user positions with high precision. The sensor fusion module employs a dynamic weighted averaging scheme to combine the heterogeneous sensor and signal inputs, assigning adaptive reliability weights to each source depending on environmental context (e.g., indoor, or outdoor conditions). The methodology was implemented and evaluated using MATLAB and Simulink in both simulated and real‐world testbeds, encompassing indoor and outdoor scenarios with varying interference levels and mobility patterns. Experimental results demonstrate a mean localization accuracy (RMSE) of 0.85 m indoors, outperforming Wi‐Fi‐only (2.4 m) and Bluetooth‐only (1.7 m) methods. Moreover, the proposed system maintained stable accuracy under multi‐path interference and signal fading, exhibiting 15% higher robustness than traditional techniques. The real‐time implementation achieved a response latency of 300 ms and an update rate of 5 Hz, supporting continuous user tracking in dynamic environments.