Event-triggered sequential fusion filter for nonlinear multi-sensor system with random packet dropout and composite correlated noise

This study investigates event-triggered (ET) mechanism-based sequential fusion filtering for nonlinear multi-sensor systems with composite correlated noise and random packet dropout. Composite correlated noise means that the observation noise at the same time is correlated, while the observation noise is also correlated with the process noise at the previous and the same instants. To minimize the energy loss within the system, the ET mechanism is used to reduce the communication rate of the sensors. The equation for the local filtering process is derived based on whether the ET mechanism has been triggered. To improve the accuracy of the filter, a noise estimator for local observations is developed using the ET mechanism. A two-step prediction method is used to calculate the estimate state vector of the main filter for time updating. In this paper, the sequential fusion filtering algorithm under the Bayesian framework is derived to solve the nonlinear systems with above problems, and the numerical implementation of the algorithm is given by the square-root cubature Kalman filter (SCKF). Finally, simulation experiments demonstrate that the developed algorithm enables a reduced communication rate of the sensor system while maintaining powerful filter estimation performance.