Globally Optimal Sequential Fusion Filtering for Multi-Sensor Systems with DoS Attacks and Correlated Noises

This paper is concerned with the globally optimal sequential fusion estimation problem for multi-sensor stochastic systems subject to denial of service (DoS) attacks and correlated noises. The stochastic uncertainties existing in the state model are represented by correlated multiplicative noises. The additive noises in the system and observation equations are correlated at the same time stamp. Then, the prediction compensation strategy for packet losses caused by DoS attacks is selected and a new equivalent system model with virtual noise is established. Based on the new system model and an innovative analysis approach, an optimal centralized fusion filter is proposed in the linear minimum variance sense. To reduce computational cost, a globally optimal sequential fusion filter is presented, which requires the calculation of the estimators for system and observation noises. It has the same estimation accuracy with the centralized fusion filter. Finally, an illustrative example verifies the effectiveness of the proposed algorithms.