I have no background in control theory but this sounds very similar to the identifiability problem in nonlinear ICA. Are those equivalent?