Framework for reliable and automatic multi-frequency correction of non-anechoic antenna measurements

• Antenna measurements are normally performed in expensive laboratories. • Acceptable data can also be obtained in non-anechoic conditions upon refinement. • Correction performance depends on manual setup of post-processing algorithms. • A deterministic framework for automatic tuning of post-processing setup is proposed. • Method adjusts correction setup to frequency, antenna, and propagation conditions. Correction of antenna far-field measurements performed in uncontrolled environments is a challenging problem. Non-anechoic sites, while cheap compared to professional laboratories, suffer from poor propagation properties due to multipath interferences and noise from external sources of electromagnetic radiation. Direct measurements in such conditions are ineffective for drawing meaningful conclusions on antenna performance. The quality of obtained responses can be significantly enhanced through application of appropriate post-processing approaches. Unfortunately, performance of available techniques is subject to experience-based adjustment of algorithm-specific setup, often aided with rule-of-thumb solutions determined based on rudimentary experiments. In this work, limitations of existing manual, or semi-manual approaches to post-processing setup are showcased in contrast with an unsupervised tuning that employs rigorous numerical optimization. The proposed framework automates the determination of correction settings based on analysis of discrepancy between the refined antenna measurements against the responses obtained from electromagnetic simulations. It is the first deterministic, numerically-driven tool tailored for post-processing of data obtained in non-anechoic environments while eliminating the need for cognitive experimentation. A thorough evaluation of the method was performed in two test sites, incorporating five different setups and five correction techniques, for a total of 1250 experiments executed across 25 independent frequencies of interest and three antenna structures. Benchmark results that demonstrate usefulness of automatic setup adjustment when compared to rule-of-thumb methods are also provided.