ON SOME ASYMPTOTIC UNCERTAINTY BOUNDS IN RECURSIVE LEAST-SQUARES IDENTIFICATION

This note deals with the performance of the recursive least squares al gorithm when it is applied to problems where the measured signal is co rrupted by bounded noise. Using ideas from bounding ellipsoid algorith ms we derive an asymptotic expression for the bound on the uncertainty of the parameter estimate for a simple choice of design variables. Th is bound is also transformed to a bound on the uncertainty of the tran sfer function estimate.