Overlapping Fusion Estimation for Discrete Time-Varying Interconnected Systems

This letter is concerned with the distributed estimation problem for discrete time-varying interconnected systems with overlapping states. A fusion strategy, which weights neighbors' overlapping state estimates into a better result, is used to exploit the redundant information from sensors. Under this fusion strategy, an overlapping fusion estimator is proposed with only the information from each subsystem and its neighbors. By the idea of constructing and minimizing an upper bound of the square error of the estimate, convex optimization problems are established in terms of linear matrix inequalities to obtain the fusion weights and estimator gains successively. Notice that these optimization problems can be directly solved by standard software packages, and the computational complexity of each optimization problem is low, even for large-scale systems. Moreover, conditions on coupling structures are provided to guarantee the stability of the overlapping fusion estimator. Finally, an illustrative example is employed to show the effectiveness of the proposed methods.