Fast and accurate database using N-linear interpolation for reflectarray analysis, layout design and crosspolar optimization
Subject:
Database
look-up table (LUT)
N-linear interpolation
machine learning
surrogate model
support vector regression (SVR)
crosspolar optimization
reflectarray antenna
Publication date:
Descripción física:
Abstract:
This work provides an assessment in terms of computational speed and accuracy at the radiation pattern level of a database to characterize the reflectarray unit cell for use in analysis, layout design and crosspolar pattern optimization. Details of an efficient implementation of the database are provided, and its performance is compared to that provided by a machine learning technique based on support vector regression (SVR). A method of moments based on local periodicity serves as the baseline for the comparison, both in terms of accuracy and acceleration. The results show that, even when the database uses a simple N-linear interpolation, accuracy is similar to that provided by the SVR technique while accelerating the analysis and layout design of reflectarray antennas.
This work provides an assessment in terms of computational speed and accuracy at the radiation pattern level of a database to characterize the reflectarray unit cell for use in analysis, layout design and crosspolar pattern optimization. Details of an efficient implementation of the database are provided, and its performance is compared to that provided by a machine learning technique based on support vector regression (SVR). A method of moments based on local periodicity serves as the baseline for the comparison, both in terms of accuracy and acceleration. The results show that, even when the database uses a simple N-linear interpolation, accuracy is similar to that provided by the SVR technique while accelerating the analysis and layout design of reflectarray antennas.
Description:
European Conference on Antennas and Propagation, EuCAP 2022 (16th. 2022. Madrid, Spain)
Patrocinado por:
This work was supported in part by the Ministerio de Ciencia, Innovación y Universidades under project IJC2018-035696-I; by the Ministerio de Ciencia e Innovación and the Agencia Estatal de Investigación within project ENHANCE-5G (PID2020-114172RB-C21 / AEI / 10.13039/501100011033).