Information fusion based filtering for multi-sensor system

In allusion to the state estimation problem of the multi-sensor system, two filtering algorithms based on information fusion estimation theory are presented, which called the measurement fusion filtering and the state fusion filtering. The former is based on the idea of fusion first and then filtering. It fuses the sub-systems measurements information to obtain the system measurement estimation, and then fuses the system state predictive information to obtain the state estimation. The latter is based on the idea of filtering first and then fusion. It fuses the predictive information and the measurement information of the sub-systems states to obtain the sub-systems states estimation, and then fuses all sub-systems states estimation information to obtain the system state estimation. The former filtering is same with the centralized fusion filtering, while the latter filtering is different, because of the different fusion information. The performance of the proposed filtering methods depends on the utilized information weight.