Asymptotic efficiency of manifold ambiguity resolution for DOA estimation in nonuniform linear antenna arrays

Linear dependence amongst the nonuniform linear array antenna manifold ("steering vectors") leads to a failure of the MUSIC algorithm to identify the correct signal sources (a "manifold ambiguity"). However, this failure does not necessarily imply that unambiguous DOA estimates are unobtainable. For fully-augmentable array geometries, an algorithm involving the standard augmentation approach successfully resolves the manifold ambiguity. For partially-augmentable and arbitrary-geometry arrays, the augmentation approach was generalised, mainly relying upon maximum-entropy covariance matrix completion. This paper deals with an asymptotic analysis of the generalised approach which is capable of resolving manifold ambiguity (provided the corresponding Fisher matrix is not rank-deficient).