Study of Super-resolution Direction Detection with Spatially Nonstationary Noise

This paper studied Super-resolution Direction Detection of arbitrary spatial sensors array, with nonstationary spatial noise, i.e. a noise has a diagonal covariance matrix whose diagonal elements have unequal noise power. An improvement of covariance matrix difference approach was made. The proposed algorithm conducts the difference between the averaging covariance matrix and its transform matrix to eliminate the noise effects which can improve the performance of DOA estimation. Gerschgorin Radii is applied to super-resolution direction detection algorithm, which can exactly estimate source number, so improve precision of super-resolution direction detection. This method conducts covariance matrix transform, so segregate noise disks from signal disks, then estimate using Gerschgorin Disks. Computer simulation is conducted under circumstance of arbitrary spatial sensors array, and compare with MDL Principle and MUSIC algorithm. The result shows the methods is useful for super-resolution direction detection with nonstationary spatial noise, and improve precision of direction estimation.