Distributed Reduced-order Fusion Wiener Filter for Descriptor systems

Based on the weighted fusion algorithms in the linear minimum variance sense and canonical form of descriptor systems, the distributed optimal weighted reduced-order fusion Wiener filters are given for descriptor discrete-time stochastic linear systems with multiple sensors. They have decoupled structure for state and asymptotic stability. Further, they have the reduced online computational burden since steady-state values of local gain matrices, covariance matrices and fusion weights can be computed offline. A simulation example verifies the effectiveness.