Distributed Maneuvering Vehicle Tracking Algorithm Using Ultra-wideband in Convoluted Indoor Environments

Tracking maneuvering vehicles via ultra-widebandbased wireless sensor networks is challenged by non-stationary, heavy-tailed noise and intermittent measurements. We therefore propose a distributed tracking algorithm tailored for nonstationary, non-Gaussian, and nonlinear environments. First, we introduce a comprehensive system model to capture these complexities. We then employ the minimum error entropy (MEE) criterion, variational Bayesian (VB) inference, and the unscented transform to suppress non-Gaussian noise, infer noise statistics online, and handle nonlinear dynamics. Building on this, we develop an interactive multiple-model Kalman filter—dubbed IMM-DMEE-UKF-VB—and rigorously analyze its mean performance and consistency. Finally, extensive simulations and experiments confirm the algorithm's effectiveness.