The blind identification of vector-valued FIR transfer functions using excl
usively second-order statistics of the output has become conventional. The
so-called subspace method provides a nice and efficient way of solving the
problem, However, the extension to the multi-input/multi-output (MIMO) case
is not that straightforward, and, for instance, some indetermination drawb
acks may occur. We show in this correspondence that the introduction of som
e cyclostationarity at the input (which still leads to feasible systems) ma
y lead to an algorithm that is far more satisfactory because for both, it i
s less prone to indetermination problems, and the required assumptions are
not too sophisticated.