Algebraic parameter estimation of damped exponentials

The parameter estimation of a sum of exponentials or the exponential fitting of data is a well known problem with a rich history. It is a nonlinear problem which presents several difficulties as the ill-conditioning when roots have close values and the order of the estimated parameters, among others. One of the best existing methods is the modified Prony algorithm [1] which suffers in the presence of noise. In this paper we propose an algebraic method for the parameter estimation. The method, differently from the modified Prony method, is considerably robust to noise. The comparison of both through simulations confirm the good performance of the algebraic method.