Comments on 'On the estimation of continuous-time transfer functions' by L. Wang and P. Gawthrop

The stimulating paper by Wang and Gawthrop (2001: WG from hereon) presents some interesting new results concerning the identi®cation and estimation{ of continuous-time, linear Transfer Function (TF) models. The purpose of this note is to point out that alternative Instrumental Variable (IV) methods have been available for some considerable time and seem to satisfy the requirements speci®ed by WG as regards the identi®cation of model order and optimality. The e cacy of these alternative methods, which are available in the CAPTAIN Matlab Toolbox (http://www.es.lancs.ac. uk/cres/captain/), is demonstrated on the simulation example used by WG. The results show that the optimized pre®lters, that are so important in continuous-time estimation, are more precisely de®ned in the IV case, leading to superior estimation results.