Adaptive Superresolution Phenomena In Aberrant Arrays
N.K. JabIon, Bernard Carlos Widrow · 2005
Maximum-likelihood (ML) beamformers can perform accurate bearing estimation by taking advantage of the fact that for high input signal-to- noise ratios, the adaptive beamwidth is much nar- rower than the classical Rayleigh limit. This paper studies the effect of random antenna element errors on this %uperresolution phenomenon. The imper- fections considered are amplitude and phase errors which are constant during the period of adaptation. With a single input point-source signal, it is shown that the ML beamformer half-power adaptive beamwidth is degraded by array aberrations, but it is still several times better than that of conventional nonadaptive beamformers, provided the input signal-to-noise ratio is 0 dB or higher. A simple scalar formula is presented for the ML half-power beamwidth in the presence of element errors for uni- formly spaced line arrays, and simulations are described which verified its accuracy to within 4%. Lastly, two new on-line algorithms are proposed to make the superresolution concept robust to array imperfecti 08s.