Abstract: | The MUSIC estimator of two-dimensional frequencies (2-D MUSIC) is studied assuming a one-measurement data model with deterministic phases and additive complex white Gaussian noise. The large sample estimation covariance of the 2-D MUSIC is derived and compared to that of the 2-D matrix pencil (MP) estimator. The theoretical estimation variances for both the MP and MUSIC estimators are compared with the simulated MP and MUSIC estimation variances and the Cramer-Rao Bound (CRB). In the single 2-D sinusoid case, the most revealing form of the estimation covariance for both estimators are provided. The results shown in this paper are valid for a median range of SNR. |