Fusion and Comparative Study of Fuzzy k-Nearest Neighbors, Random Forest, and Gradient Boosting for Indoor Positioning Systems via WLAN/WiFi Fingerprinting

This paper proposes a novel indoor positioning system based on WLAN/WiFi fingerprinting leveraging a multi-model fusion approach. We combine the strengths of Fuzzy k-Nearest Neighbors (FKNN), Random Forest (RF), and Gradient Boosting (GB) algorithms to achieve superior positioning accuracy. Each model is meticulously implemented and optimized using the publicly available UJIIndoorLoc dataset, a comprehensive benchmark for indoor localization. Our experimental results demonstrate that the fusion of these models significantly outperforms individual approaches, achieving substantial improvements in indoor positioning accuracy.