Efficient Shaped-Beam Reflectarray Design Using Machine Learning Techniques

Abstract

This papers introduces the use of machine learning techniques for an efficient design of shaped-beam reflectarrays considerably accelerating the overall process while providing accurate results. The technique is based on the use of Support Vector Machines (SVMs) for the characterization of the reflection coefficient matrix, which provides an efficient way for deriving the scattering parameters associated with the unit cell dimensions. In this way, the SVMs are used within the design process to obtain a reflectarray layout instead of a Full-Wave analysis tool based on Local Periodicity (FW-LP). The accuracy of the SVMs is assessed and the influence of the discretization of the angle of incidence is studied. Finally, a considerable acceleration is achieved with regard to the FW-LP and other works in the literature employing Artificial Neural Networks.

Artificial intelligence for satellite communication: A review

“…For example, NNs were used by Shan et al [203] to forecast the phase-shift, whereas kriging was suggested to forecast the electromagnetic response of reflectarray components [204] . Support vector regression (SVR) has been used to accelerate the examination [205] and to directly optimize narrowband reflectarrays [206] . To hasten calculations without reducing their precision, Prado et al [207] proposed a wideband SVR-based reflectarray design method, and demonstrated its ability to obtain wideband, dual-linear polarized, and shapedbeam reflectarrays for direct broadcast satellite applications.…”