Gaussian Belief Propagation-Enhanced CRB-KF Framework for Cooperative Localization

This paper proposes a Gaussian belief propagationenhanced Cramér-Rao bound Kalman filter (GBP-CKF) framework to address cooperative localization (CL) in vehicular ad hoc networks (VANETs). The framework combines the strengths of CRB-KF for nonlinear, non-Gaussian systems with the distributed capabilities of Gaussian belief propagation to improve localization accuracy and fault tolerance. We construct a factor graph integrating odometry data and intervehicle observations, and perform message passing for vehicle state estimation to ensure robust operation along factor graph edges, optimizing error distribution and enhancing localization performance. Simulation results demonstrate that our method significantly outperforms conventional Kalman filters and factor graph optimization approaches in both localization accuracy and computational efficiency.