Spatial and temporal frequency estimation of uncorrelated signals using subspace fitting

We present a novel method for spatial and temporal frequency estimation in the case of uncorrelated sources. By imposing the diagonal structure given in the signal covariance matrix, it is possible to improve the performance of subspace based estimators. The proposed method combines ideas from subspace and covariance matching methods to yield a non-iterative frequency estimation algorithm. In a numerical example we show that the estimator has a lower small sample resolution threshold than root-MUSIC and similar large sample performance