Multisensor fusion for state estimation of linear models in the presence of bounded disturbances

A hierarchical state bounding estimation method is presented for a linear dynamic system where different sensors offer several measurements of the same state vector, each of which is subject to unknown but bounded disturbances and is equipped with a local processor. The proposed algorithm works at two levels : at each sampling time, each local processor computes the state estimate and its outer-bounding ellipsoid according to the local measurements given by the corresponding sensor. This results are transmitted simultaneously from all local processors to the fusion center which computes the global estimate and its bounding ellipsoid, then it feeds this information back to all the local processors. This feedback allows the local processors to adjust their results by taking into account the measurements of all the other sensors.