Analysis of the MUSIC estimator for the detection of frequencies in sinusoidal signals
Résumé
The paper considers the application of the subspace method MUSIC to the estimation of rolling, nutation and precession frequencies of a high velocity spinning artillery shell. The influence of the various parameters such as signal-to-noise ratio and slow drift of frequencies is studied. The formula giving the variance of the estimation is detailed at the light of the specificities of the application, and a novel result is established stating that when the soughtafter frequency is varying in a time-affine manner over the considered data window, the MUSIC estimate recovers the average value of the frequency. A typical case-study serves as illustration, assessing that, at the expense of a increase of the computational load, MUSIC can recover the nutation frequency of the shell with a standard deviation below 1 rad/s.
Origine : Fichiers produits par l'(les) auteur(s)