Continuous-time frequency domain subspace system identification

In this paper we present a new subspace identification algorithm for the identification of multi-input multi-output linear time-invariant continuous-time systems from measured frequency response data. We show how the conditioning of the data-matrices in the algorithm can be improved by making use of recursions derived from the Forsythe polynomials. The asymptotic properties are analyzed and it is shown that, when the error distribution on the measurements is given, the algorithm can be made asymptotically unbiased through the introduction of a weighting matrix.