A Fast Algorithm for Errors-in-Variables Filtering

This note concerns the optimal estimation of the input and output sequences of linear time-invariant errors-in-variables (EIV) processes. An efficient recursive filtering algorithm is proposed. It is an innovation-based approach that relies on the triangular decomposition of block Toeplitz matrices. Unlike the other algorithms described in the literature, the proposed one is characterized by a computational complexity which increases only linearly with the order of the process. Both the SISO and MIMO cases are analyzed. An extension of the described algorithm to EIV models with colored input and output noises is considered as well.