Two-level robust sequential covariance intersection fusion Kalman predictors over clustering sensor networks with uncertain noise variances

This paper studies the problem of designing two-level robust sequential covariance intersection SCI fusion Kalman predictors for the clustering sensor networks with noise variances uncertainties. The sensor networks consist of many clusters, which are partitioned by the nearest neighbour rule. According to the minimax robust estimation principle, based on the worst-case conservative clustering sensor network with the conservative upper bound of noise variances, the two-level SCI fusion Kalman predictors are presented where the first level is the local SCI fusion predictors and the second level is the global SCI fusion predictor. This two-level fused structure can significantly reduce the communicational burden and save the energy sources. The robustness of the local and fused Kalman predictors is proved based on the Lyapunov equation method, and the robust accuracy relations are proved. A simulation example verifies the correctness and effectiveness of the proposed robust SCI predictor.