Effects of random sensor motion on bearing estimation by the MUSIC algorithm

Abstract

The MUSIC method of bearing estimation is a high resolution estimating method utilising the orthogonal relationship between the signal and noise subspaces which are spanned by the eigenvectors of the covariance or the crossspectral density matrices. In the application of the MUSIC method to bearing estimation in a sonar environment, an important source of error is that the positions of the array of sensors are under constant perturbation. The paper investigates the effect of this source of error on the bearing estimate. The Cramér-Rao bound of bearing estimation under the influence of this source of error is evaluated and the performance of the MUSIC method is compared with this lower bound.