Range Extension in Partial Spherical Near-Field Measurement Using Machine Learning Algorithm
Rezvan Rafiee Alavi, Rashid Mirzavand
Abstract
Often due to physical limitations, there is a gap in the very-near-field and near-field (NF) measurements of antennas. However, to compute the complete far-field (FF) pattern, near-field data over the whole measurement sphere are required. In this letter, an iterative extrapolation-based machine learning algorithm is presented to expand the region over which the calculated far-field is more accurate. In each iteration, the well-known analysis of variance test is used to check the overall feasibility of the regression model and derive the coefficients of the extrapolation function. To validate the method, three examples with a folded dipole antenna at 1 GHz, a vivaldi antenna at 5 GHz and a dual-frequency planar antenna are presented using both simulated and measured full and truncated near-field data.