Simulation-driven size reduction of antenna structures using adjoint sensitivities and trust regions

Explicit size reduction of antenna structures using cost efficient EM-simulation-driven optimization is discussed. The proposed approach directly handles the antenna size (as the main design objective) while ensuring satisfaction of electrical performance parameters by means of suitably defined penalty functions. For the sake of computational efficiency, optimization is carried out using cheap adjoint sensitivities and trust region framework utilized as convergence safeguard. Our technique is illustrated through a design of a compact quasi-isotropic dielectric resonator antenna.