Consistent fusion of correlated data sources

The selection of the appropriate fusion algorithm depends on the underlying data fusion architecture. In the centralized scheme, the sources are known to be independent and the Kalman filter provides an optimal solution. Unfortunately, in the decentralized architecture, the sources become correlated and the Kalman filter cannot be applied. The covariance intersection method has been proposed as a solution to the problem of decentralized data fusion. However, it results in a decrease in performance. A new fusion algorithm (largest ellipsoid approach) that avoids both of the inconsistency of the Kalman filter and the lack of performance of the covariance intersection is proposed. The superiority of the proposed approach is illustrated using the target´s tracking problem.