Reduced complexity covariance matrix estimate for subspace-based array processing

In subspace based DOA algorithms estimating the signal or noise subspace accurately, is imperative as it is the foundation for which all such algorithms are built upon. Estimating the subspaces is achieved from decomposing a data matrix or the spatial covariance matrix each incurring a computational burden. We propose to reduce the computational complexity of estimating the noise subspace by using a computationally efficient covariance matrix estimate, whose multiplications are independent of data size. We investigate the effects of the non-linearity on the noise subspaces and to the DOA estimation using the MUSIC algorithm.