Fisher Information-Based Adaptive Sampling Strategy for Near-Field Far-Field Transformation of Parabolic Antenna

Abstract

This study proposed an efficient near-field sampling strategy based on Fisher information to predict the far-field radiation pattern with a small number of near-field data. To measure the far-field radiation pattern of large antennas, such as parabolic antennas, conventional far-field measurements are often limited by spatial and monetary constraints. As an alternative, near-field to far-field transformation (NFFFT) is widely used with measured near-field data. Nevertheless, high-frequency large parabolic antennas require wide areas and fine spatial resolution of near-field measurement, making full-area scanning time-consuming and expensive. In practical settings where rapid performance verification is needed, an efficient alternative to full near-field measurements is essential. The proposed method initially models the near-field distribution mathematically using a function that combines sinusoidal and Gaussian mixture models (GMMs). Then, Fisher information is suggested to compute information at each location to prioritize the next sampling points. Numerical simulations and experimental measurements demonstrate that the model-based adaptive sampling strategy effectively reduces the number of measurement points required, while still enabling accurate prediction of essential radiation characteristics. The proposed method achieves excellent sampling efficiency, requiring only 2.4 % of the total near-field data and reducing probe movement distance by 38.9 %.