Fast Antenna Array Diagnosis from a Small Number of Far-Field Measurements

Abstract

The fast diagnosis of antenna arrays from a small number of far-field measurements is addressed. With the a priori knowledge of the failure-free array radiation pattern, it is possible to reformulate the diagnosis problem such as only the faulty elements or the localized field differences have to be retrieved. Efficient and readily available sparse recovery algorithms can then be applied to identify the failures from a small number of measurements compared to standard diagnosis techniques and hence speed up the diagnosis. More specifically, three regularization procedures namely the minimization of the $\ell_1$ , total variation (TV), and the mixed $\ell_1/\ell_2$ norm are used to solve the ill-posed array diagnosis problems. These approaches are compared to two standard fault identification techniques: the back-propagation algorithm (BPA) and the matrix inversion method for the diagnosis from simulated and measured data. The simulation of a $10 \times 10 $ waveguide array in realistic conditions of noise and taking into account the potential scaling factor between two measurements is first presented. Then, a reflectarray composed of 193 cells with metallic strips to emulate phase failures is considered. Both numerical and experimental results confirm the effectiveness of the sparse recovery algorithms and the importance of prior information on the source.