We consider the problem of constructing suboptimal filters (lower-order optimal
filters, i.e., filters for linear vector functionals of the system state vector) for stochastic multivariable
multicriteria plants. The method for constructing such filters is presented in the
Luenberger canonical basis. Using a numerical example of a seventh-order system, we show
that the proposed approach improves the optimality of filters compared to filters based on
scalar observers.