Self-tuning measurement fusion Wiener filter for autoregressive signals

Abstract

For the autoregressive (AR) signals with multisensor, unknown model parameters and unknown noise variances, using the recursive extended least square (RELS) and the correlation method, the strong consistent information fusion estimators of model parameters and noise variances are presented, and then by substituting them into the optimal weighted measurement fusion Wiener filter based on the autoregressive moving average (ARMA) innovation model, a self-tuning weighted measurement fusion Wiener signal filter is presented. Further, applying the dynamic error system analysis (DESA) method, it is proved that the self-tuning fused Wiener filter converges to the optimal fused Wiener filter in a realization, so that it has asymptotically global optimality. A simulation example shows its effectiveness.