A new dynamic vector formed information sharing algorithm in federated filter

The filterʼs fault tolerance and precision are directly affected by the information sharing principle in a federated filter. Normally the scalar-form information sharing coefficient cannot fully reflect the differences of each state variable of each filtering subsystem and track their changes. In order to overcome this shortage, a new dynamic vector-form information sharing algorithm using the eigenvalue of error covariance matrix and the singular value of observability matrix is proposed. Firstly, after presenting the conventional federated filter algorithm, the shortage of traditional scalar-form information sharing coefficient is analyzed. Secondly, with the matrix of error covariance and observability of the system, this paper designs a vector-form information sharing algorithm so that each state variable can get a different coefficient. Then, this vector-form information sharing algorithm is improved to solve the problem of asymmetry of error covariance matrix. Lastly, simulation results show that this new algorithm can help the federated filter system reasonably feedback the global information according to the information quality and observability of each sensor within the process of navigation, so that the precision of federated filter can be improved effectively in a high dynamic environment.