Using the modern time series analysis method, based on the autoregressive m
oving average (ARMA) innovation model and white noise estimators, two time-
domain approaches to multichannel optimal deconvolution are presented. In t
he first approach, the multichannel optimal deconvolution estimators are gi
ven in the ARMA innovation filters form, where the solution of the Diophant
ine equations is required. Their global and local asymptotic stability is p
roved. In the second approach, the multichannel ARMA recursive Wiener decon
volution filters without the Diophantine equations are presented, which hav
e asymptotic stability. The relationship between the ARMA innovation filter
s and ARMA Wiener deconvolution filters is discussed. Each approach can han
del the deconvolution filtering, smoothing and prediction problems in a uni
fied framework. An illustrative example and two simulation examples show th
eir effectiveness.