Event-triggered distributed diffusion robust nonlinear filter for sensor networks

Abstract

This paper focuses on the issue of event-triggered nonlinear state estimation for multi-sensor networks. An event-triggered mechanism reduces data transmission, balancing communication rate and estimation performance through triggered thresholds. After that, a novel event-triggered robust filter is proposed. The non-triggered case is a non-Gaussian process. The fading matrix adaptively adjusts the noise variance and the gain matrix is designed by the maximum correntropy criterion, avoiding the conservatism and randomness brought by the upper bound. Subsequently, an event-triggered distributed diffusion robust cubature Kalman filter is presented relying on the cubature criterion, covariance intersection technique and diffusion fusion strategy. Compared with average consensus fusion, the error covariance is utilized to compute the weights in real time and does not involve complicated iterative processes. Moreover, the consistency, convergence and stability are proven under certain conditions. Finally, the simulation results verify the effectiveness and accuracies of the proposed algorithm.