Rapid Microwave Design Optimization Using Variable-Resolution Electromagnetic Simulations

Abstract

Optimization-based parameter tuning is a necessary yet CPU-intensive stage of microwave design process. Seeking computational speedup, this work proposes utilization of variable-resolution electromagnetic (EM) models. The optimization process starts with the lowest-fidelity representation of the structure at hand, which is gradually refined to conclude the run at the high-fidelity level. The resolution adjustments involve the algorithm convergence status and are demonstrated to lead to over sixty percent cost reduction with little impact on the design quality.