A Discriminant-based RMSE Improvement Technique for Classical Prony Method in Small Array Radars

A simple DoA estimation technique that virtually reduces the noise sensitivity of the classical Prony method is presented. With the discriminant test that detects the potential outliers, the proposed scheme provides a better RMSE in the classical Prony DoA estimation with a low computational cost. Unlike the traditional enhancement techniques for classical Prony, where a priori order information is necessary for the proper operations, the proposed technique works without such explicit information. The discriminant uses a predetermined threshold value, which is chosen to effectively eliminate the undesirable outliers in the 8-element uniform array. Simulations show that the RMSE reduces to the close level of LS-Prony characteristics, while some of the estimation results are discarded by the discriminant test.