The effects of nonuniform and correlated noise on superresolution algorithms

The effects of nonuniform noise power and correlated noise in an antenna array on the performance of the recursive suppression, standard MUSIC, and reduced MUSIC superresolution algorithms are examined. It is shown that the speeds of resolution of all the algorithms are slowed, but their steady-state errors exhibit only a mild sensitivity to moderate noise-power variations if the signal-to-noise (SNR) is sufficiently large. The sensitivity of the standard MUSIC algorithm to correlated noise is shown to be much more than that of the other two algorithms, which can tolerate considerable correlated noise even when the SNR is relatively low.