Data Fusion of Unknown Correlations using Internal Ellipsoidal Approximation

To avoid both the inconsistency of the Kalman filter and the performance conservation of the covariance intersection (CI) in the case of unknown correlations, an internal ellipsoidal approximation (IEA) method is proposed to fuse the local estimations. A numerical example of three-state radar tracking system is presented to illustrate the implementation and effectiveness of the proposed algorithm. From the simulation results in the cases that the sources are ( ) independent; ( ) correlated and the cross i ii covariance are exact known; and ( ) correlated iii with unknown cross-covariance, it is obvious to see that the IEA method, like CI, circumvents the need for prior knowledge of the correlations but it gets better fusion accuracy than CI.