Impedance Database Generation for the Automated Design of Metasurface Antennas

Metasurface antennas are thin two dimensional metamaterial layers that allow or inhibit the propagation of electromagnetic waves in desired directions in free space or antenna surface. In literature they have been demonstrated to guide and make surface waves radiate to obtain a specific desired radiation pattern. Their low cost, easy integration and low losses make them promising candidates for implementing solutions of next generation receivers and transmitters for high speed wireless telecommunications, in particular for 5G and beyond-5G standards. These antennas are planar structures made up by sub wavelength periodic conductive elements, printed on a grounded dielectric substrate. By changing the individual elements shape it is possible to obtain a given current distribution, thereby design a suitable radiation pattern. However, the analysis and design of such structures is particularly challenging because of the multi-scale features that characterize the patterned surface: electromagnetic behavior is usually described macroscopically by a simplified model with the Impedance Boundary Conditions (IBC). The antenna design is usually done in two separate steps. First the IBC profile is obtained, then the individual cell shapes are extracted from a pre-computed database. In literature are already present several numerical methods aimed at automating the design of these IBC profiles, starting from the antenna domain and the constraints on the radiated field; much works remains to be done in the efficient computation of the database of the individual cells in order to properly cover the desired IBC values. In general, each cell shape is defined by a set of parameters and it realizes a certain impedance value that changes deeply with the parameters combination. The main objective of this work is to create a database of different reference cell geometries (e.g. circle, ellipse, etc.), at a single or multiple frequencies, in such a way that the resulting impedance values are distributed around the desired ones or as uniform as possible in the domain of existence. The software developed in this thesis is able to take a generic user defined cell geometry and its parametrization, evaluate the impedance value from the scattered fields obtained from simulations, and automatically choose the sampling of the parameter values in order to minimize the distance between desired impedance points. The algorithm works by partitioning the domain of interest into boxes from a coarse database of impedance points, evaluated from uniform samples of the parameters. New parameters points are taken through local interpolant inversion of the points already present. These points are then simulated with an external software module that is able to evaluate the scattering matrix of the metasurface from which it is possible to extract the new impedance points. The database is updated iteratively in this way. After exploring all boxes, it is possible to iterate again these steps in order to refine the database if needed by imposed constraints, such as the minimum number of impedance points needed. It is also possible give specific impedance points to the algorithm in order to refine the search close to these points. Numerical results highlight an improvement in terms of lower average distances between the desired points and the database points with respect older solutions. As a consequence, it is possible to realize more accurately the wanted IBC on the surface, leading to lower discrepancies in radiation pattern.