IMPROVED DETECTION OF A RANDOM AMPLITUDE SINUSOID BY CONSTRAINED LEAST-SQUARES TECHNIQUE

In this paper, we address the problem of estimating the frequency of a sinusoidal signal with a lowpass autoregressive envelope. Due to the ARMA-like form of the spectrum, the frequency is estimated by solving a system of Yule-Walker equations. In contrast to a method we previous ly proposed, the parameters are here constrained to be of a specific f orm. A Gauss-Newton algorithm is derived to solve the constrained leas t-squares problem. Simulation results show that an improvement can be achieved compared to the unconstrained method.