Feasible space boundary search for improved optimisation-based miniaturisation of antenna structures

This study describes an enhanced procedure for electromagnetic-simulation-based miniaturisation of antenna structures. The proposed methodology utilises formulation of the antenna design as non-linear minimisation problem oriented towards explicit size reduction with constraints imposed on the reflection response of the structure at hand. The primary idea is to explore the boundary of the feasible region so as to allow efficient identification of a constrained optimum that ensures satisfaction of the prescribed matching requirements for the structure at hand. The proposed algorithm overcomes the issues related to the standard formulation involving a penalty function approach, where high non-linearity of the objective function in the vicinity of the optimum may lead to finding a sub-optimal design. For the sake of demonstration and verification, three examples of ultra-wideband antennas are considered. In all cases, the proposed methodology leads to smaller footprints and better control of the in-band matching compared to the benchmark algorithm, i.e. gradient-based search with implicit reflection response control through a penalty function approach.