Optimal Component Fusion Steady-State Smoothing for Discrete Multichannel ARMA Signals

Abstract

Based on white noise estimators and the optimal fusion algorithm in the LMV (linear minimum variance) sense, distributed optimal fusion steady-state smoothers with scalar weights are given for all components of discrete multichannel ARMA (autoregressive moving average) signals with correlated noises. The precision of the fusion smoothers is higher than that of local smoothers, but is lower than that of the fusion smoother with matrix weights. However, the computational burden can be reduced since only scalar weights are required. Applying it to a double-channel ARMA signal system with three sensors shows the effectiveness