Identifiability of multivariable dynamic errors-in-variables systems

Research in the area of identifiability of dynamic Errors-in-Variables (EIV) systems (i.e., systems where both their inputs and outputs are measured with errors) has been substantial in the last decades. However, most of the literature has focused on Single-Input-Single-Output (SISO) systems. In this paper we extend recent identifiability results to multivariable (MIMO) systems. To this end, we use some results on the canonical factorization of matrix functions. In particular, we establish generic global identifiability for EIV models up to a scalar constant, which is surprising, since it corresponds exactly to what happens in the SISO case. We then present several sets of conditions, analogous to those usually assumed for SISO systems, to ensure full global identifiability.