Comparison of Distributed Fusion Filters for Linear Dynamic System with Uncertainty

In this paper, a distributed fusion filtering problem for a linear discrete-time dynamic system with uncertainty is considered. All fusion filtering algorithms are based on fusion formulas which represent a weighted sum of the local Kalman estimates with matrix or scalar weights. The fusion weights are calculated by using four algorithms: convex combination, optimal fusion, covariance intersection, and median fusion. The comparison results of the fusion algorithms are discussed in terms of estimation accuracy and computation cost.