On combining the eigenstructure-based and model-based approaches in direction of arrival estimation problems

The authors consider the problem of estimating direction-of-arrival (DOA) parameters from observations made by an array of sensors in a linear equispaced antenna array operating at narrowband. They propose an approach that involves application of model-based autoregressive methods to the signal-space eigenvectors found from a decomposition of the sample covariance matrix. The rationale for the approach is presented. Preliminary simulation results suggest that the method may have advantages over MUSIC at low (5 dB) SNR