Nonlinear distributed estimation fusion that reduces mean square error

This paper considers distributed estimation in multisensor tracking systems with and without knowledge about cross-covariance matrices among the local estimation errors. Nonlinear fusion rules are proposed to reduce the mean square error (MSE) of the estimate. Based on the best linear unbiased estimation fusion and covariance intersection fusion formulas, several classes of nonlinear estimators are proposed, which have a lower MSE than existing linear unbiased fusers. Some numerical examples are provided to verify the theoretical analysis and to illustrate the performance of the proposed estimators. Keywords: Distributed fusion, nonlinear estimation, mean square error, least squares.