Polyhedric approximation and tracking for bounded-error models

Limited-complexity polyhedric methods are presented for estimating the parameters of a linear model from bounded-error data. The simplest polyhedric description that can be obtained is under the form of a simplex, but polyhedra with more faces can be considered as well. All descriptions are guaranteed to contain the posterior feasible set for the parameters. A new polyhedric algorithm is suggested for tracking time-varying parameters. it can accommodate both smooth and infrequent abrupt variations of the parameters. The complexity of the polyhedric tracker can also be limited