Cognition-Driven Formulation of Space Mapping for Reducing Gain Variation of Antennas

The cognition-driven formulation of space mapping (SM) is effective for equal-ripple optimization of microwave filter. In this paper, we propose to use cognition-driven SM to reduce the gain variation of antennas. After the design process of an antenna, the gain at different frequencies can be different. To achieve a better performance, the gain variation at different frequencies should be reduced. In this paper, gain values at different frequencies of antennas are used as feature parameters. We build a mapping from the feature parameters to the design variables. With mappings from the design variable space to feature parameter spaces, we can reduce the gain variation. The trust region method is used to guarantee the convergence of this process. A Yagi-Uda antenna is used to illustrate our proposed method