The antenna characterization from planar near-field (NF) measurements is generally realized by using the classical NF to far-field transform technique of plane wave expansion (PWE). This approach imposes strong constraints on NF sampling. To overcome these limitations, an equivalent model of the antenna under test (AUT) is created based on a distribution of infinitesimal dipoles. A reduced-order model (ROM) of the problem is constructed to obtain a decomposition basis defining the radiated field. The powerful ability of the ROM in determining the number of points needed for accurate NF measurements is demonstrated. Also, efficient non-conventional sampling strategies are applied to the case of planar NF measurements and the influence of these distributions on the reduction of the number of samples is studied. The global analysis of our approach on simulated and measured NF data shows that only 20% of the total number of points are needed with respect to the classical PWE technique to achieve an accurate characterization.
