Numerical Investigations on Phase Recovery from Phaseless Spherical Near-Field Antenna Measurements with Random Masks

Phaseless spherical near-field antenna measurements generally address the challenge of computing complex coefficients describing the antenna under test's (AUT) radiation behavior from amplitude near-field measurements. The AUT's far-field (FF) can then be obtained from those complex coefficients. Most of the techniques used in the literature result in a modified sampling method (e.g., two-spheres or sphere with two probes) and a phase retrieval algorithm (e.g., WirtingerFlow or PhaseLift). Sampling methods are chosen to increase the number of independent measurements to aid phase recovery. In our con-tribution’ we introduce the approach of random masks, leaning on the concept of diffraction patterns, which is well-known and used in the phase retrieval theory. Random masks can be seen as intentional random perturbations occurring in the measurements either at the AUT, probe or in between; or as an extension to conventional measurements to increase the number of independent measurements. State-of-the-art phaseless sampling methods can also be interpreted as masks with limited randomness. A general mathematical model is presented, and different types of masks based on random distributions are investigated through simulations on a transformation with spherical wave expansion. Firstly, generic masks are considered to benchmark the achievable reconstruction error, and secondly, masks based on probes are examined.