Distributed state estimations based on cubature Kalman filtering

This paper investigates high-dimensional distributed state estimations problems for a class of discrete-time nonlinear systems in a sensor network. We present a new distributed estimation algorithm called a novel distributed cubature Kalman filter (DCKF) based on weighted average consensus. The proposed algorithm is developed from the weighted average consensus approach and a recently developed cubature Kalman filter. Different from the existing DCKFs, the proposed filter does not require the pseudo measurement matrix during state estimations. The algorithm preserves advantages of distributed filters such as the scalability and the robustness to individual failures and has the high accuracy and strong stability of the cubature Kalman filter. Simulations illustrate the effectiveness and the superiority of the proposed distributed filter.