Total least squares techniques for high resolution direction finding

Constrained total least squares (CTLS) is a method developed to solve an overdetermined system of equations Ax=b using a generalized least-squares criterion. The CTLS technique attempts to remove the noise present in both A and b by a least-squares approach while it simultaneously exploits the algebraic structure of that noise. The technique has been applied to estimate the angle of arrival of planar narrowband wavefronts impinging on a linear uniform array. The performance of the CTLS method was better than that of the MUSIC algorithm, and its root-mean-square error is very close to the Cramer-Rao bound for medium to high SNR.<>