Feedback-Based Fusion Filtering for Rectangular Descriptor Systems via Amplify-and-Forward Relay Strategy

This paper is concerned with the feedback-based fusion filtering problem for linear rectangular descriptor systems. By utilizing full-order transformation, the original descriptor system is transformed into a full-order non-descriptor system. Based on the measurement output forwarded by the relay and the feedback information from the fusion center, the local filters are constructed, and the upper bound of each filtering error covariance matrix is obtained by using the basic inequality and mathematical induction. Moreover, the local filter gain is derived by calculating the minimum value of the obtained upper bound. The local information is comprehensively processed by using the covariance intersection fusion criterion and the corresponding feedback-based fusion filtering algorithm is proposed. Finally, a numerical simulation is given to verify the effectiveness of the designed filtering algorithm.