Distributed fusion filtering for stochastic uncertain systems subject to correlated noises, random delays and data losses

The distributed fusion filtering problem is addressed for stochastic uncertain systems with correlated noises, multi-step transmission delays and packet dropouts. Stochastic uncertainties in the state equation and measurement equation, one-step auto-correlated and cross-correlated noises as well as multi-step delays described by some Bernoulli distributed random variables are simultaneously considered. Utilizing state augmentation approach, the original system is changed into a parameterized one. The optimal local filters are then proposed by means of the innovation analysis method for each subsystem. Meanwhile, the filtering error cross-covariance matrices between any two local filters are derived. On this basis, the distributed fusion filter is designed via matrix-weighted fusion estimation criterion. Finally, the effectiveness of the designed filter is illustrated by a numerical example.