Fusion of local filters

This paper considers the problem of fusion of local filters. We derive an optimal mean square combination of arbitrary number of correlated estimates. In particular, for two sensors this combination represents the well-known Millman and Bar-Shalom-Campo formulae for uncorrelated and correlated estimation errors, respectively. The new combination is applied to an adaptive filtering problem and fusion of multisensor estimates. Two suboptimal filters with a parallel structure are herein proposed. The equation for error covariance characterizing the mean square accuracy of these filters is derived. In consequence of parallel structure of the filters, parallel computers can be used for their design. The examples demonstrate the effect of the common process noise on the fusion of the state estimates of a target based on measurements obtained by two different sensors.