Adaptive design for estimation of unknown parameters in linear systems (Q1582944)

The authors are concerned here with the design of asymptotically optimal inputs for adaptive parametric estimation in linear dynamic systems. The problem is divided into two sub-problems, in the first of which a deterministic input signal corresponding to the maximal Fisher input information is found, while in the second, asymptotically efficient estimators are obtained. To effect this, a multi-stage estimation procedure is introduced and it is shown that a Newton-Raphson type estimator may be used to correct the initial maximum likelihood estimator. Asymptotic results on optimal inputs are obtained under various assumptions as to the stability of the dynamical system.