Enhancing indoor drone localization accuracy through UWB multipath signal analysis

This paper presents a comprehensive framework for high-precision indoor drone localization by leveraging the unique properties of Ultra-Wideband (UWB) multipath signals. Traditional GPS-based systems are ineffective in indoor environments due to severe signal attenuation and non-line-of-sight (NLoS) propagation. In contrast, UWB signals offer nanosecond- level time resolution, enabling centimeter-level ranging accuracy even in the presence of multipath interference. We propose a novel Bayesian probabilistic model that transforms multipath components from nuisance parameters into valuable geometric features through the concept of virtual anchors (VAs). A factor graph-based belief propagation (BP) algorithm is developed to jointly estimate the drone’s dynamic state and the positions of both physical anchors (PAs) and VAs. Extensive simulations on a MATLAB platform demonstrate a mean localization error of 0.42 meters under NLoS conditions, outperforming existing methods by 40% . The proposed framework bridges the gap between theoretical drone localization and practical navigation applications, providing a foundation for future research on indoor drone localization.