Computationally efficient DOA estimation based on linear prediction with Capon method

Abstract

Of the several methods for estimating the direction of multiple signals with an array antenna, superresolution direction-of-arrival (DOA) estimation techniques, such as MUSIC and ESPRIT, have been in the spotlight. Although the performance of these techniques is reliable, their computational costs are considerable. We propose a new DOA estimation technique using the linear prediction (LP) method in conjunction with the Capon method. In our proposed technique, the LP method is used to estimate the true and spurious DOAs, and the true DOAs can be selected by evaluating the relative signal powers obtained by the Capon method. To estimate the number of true DOAs, we apply the values of Capon's array output power to the decision criterion, such as minimum description length (MDL). Simulation results showed that the proposed technique gives a maximum of about eighty-percent in computational cost reduction compared with MUSIC and that the technique accurately estimated the DOAs.