A robust optimization approach to kernel-based nonparametric error-in-variables identification in the presence of bounded noise

A kernel-based nonparametric approach to identification of linear systems in the presence of bounded noise affecting both input and output measurements is proposed in this paper. The problem to be solved is firstly formulated in terms of robust optimization. The solution to such a problem is then obtained by proving that the originally formulated robust optimization problem is equivalent to a standard semidefinite optimization problem (SDP) easily solvable with available software for SDP optimization (e.g., SeDuMi). The effectiveness of the proposed approach is shown by means of some simulation examples.