Abstract
Many algorithms for the optimization of array antennas need the calculation of a gradient to minimize a cost function. Usually, the best approach is to compute analytically the derivatives, since that way computations are faster. However, sometimes the derivatives are cumbersome to obtain or they cannot be calculated, such as when directly optimizing the layout in reflectarrays for cross-polarization improvement. In those cases, derivatives are evaluated using numerical techniques such as finite differences. In this work, we present the numerical technique of differential contributions (DFC) to accelerate gradient-based algorithms when the derivatives are calculated with finite differences, achieving a complexity time scaling of the same order as the analytical derivatives. The technique is applied to a far field phase-only synthesis for reflectarray antennas using the generalized Intersection Approach, and it is compared with the analytic derivatives and the use of finite differences with the FFT. The DFC technique shows superior performance in all cases, even than the analytic derivative.
Original language | English |
---|---|
Title of host publication | 2019 13th European Conference on Antennas and Propagation (EuCAP) |
Publisher | IEEE |
ISBN (Electronic) | 9788890701887 |
Publication status | Published - 20 Jun 2019 |
Event | 13th European Conference on Antennas and Propagation 2019 - Krakow, Poland Duration: 31 Mar 2019 → 5 Apr 2019 |
Conference
Conference | 13th European Conference on Antennas and Propagation 2019 |
---|---|
Abbreviated title | EuCAP 2019 |
Country/Territory | Poland |
City | Krakow |
Period | 31/03/19 → 5/04/19 |
Keywords
- array
- differential contributions
- far field
- finite differences
- generalized Intersection Approach
- Gradient-based algorithm
- optimization
- reflectarray
- synthesis
ASJC Scopus subject areas
- Computer Networks and Communications
- Instrumentation