Laboratory testing of an SVD‐based approach to recover the non‐redundant bi‐polar NF data from the positioning error affected ones

Abstract

This study deals with the problem of the correction of known probe positioning errors, which can occur in an actual planar near‐field facility when adopting a non‐redundant near‐field to far‐field (NFTFF) transformation with a bi‐polar scan for quasi‐planar antennas. To this end, a singular value decomposition based approach is developed to recover the uniform bi‐polar samples, whose position is set by the non‐redundant sampling representation got by shaping the antenna with a double bowl, from those affected by positioning errors. Then, a 2D optimal sampling interpolation formula is used to accurately reconstruct the plane rectangular near‐field data needed by the standard NFTFF transformation from the retrieved non‐redundant uniform samples. Experimental results, assessing the capability of the developed procedure to correct even pessimistic positioning errors, are shown.

Pattern Reconstruction from Near-Field Data Affected by 3D Probe Positioning Errors Collected via Planar-Wide Mesh Scanning

“…The devised procedure converges only when it is possible to formulate a bijective relationship associating every correct (regular) sampling point with the nearest improperly positioned (irregular) one. A different approach, not suffering from the above constraints, is employing the singular value decomposition procedure to compute the regular PWMS samples from those corrupted by positioning errors [21][22][23][24][25][26][27][28]. Anyhow, to avoid a huge computational effort, such an approach can be adopted only if it is possible to split the retrieving of the regular samples in two independent 1D problems.…”

“…Anyhow, to avoid a huge computational effort, such an approach can be adopted only if it is possible to split the retrieving of the regular samples in two independent 1D problems. It must be stressed that the positioning correction approaches in [18][19][20][21][22][23][24][25][26][27][28] can be applied only if the sampling points lie on the nominal measurement surface, since they are based on 2D non-redundant sampling representations. An exhaustive discussion on the reconstruction of the regular samples from the irregular ones and a related wide bibliography can be found in [18,22].…”