Genetic Algorithm-Based Inverse Design of Guided Wave Planar Terahertz Filters

TL;DR

This paper introduces a genetic algorithm-based inverse design framework for creating high-performance planar terahertz filters with specific S-parameter responses, optimized efficiently using ABCD matrix evaluation and validated with FEM simulations.

Contribution

It presents a novel GA-based inverse design method for THz filters that integrates structural connectivity constraints and accelerates optimization with ABCD matrix calculations.

Findings

Designed band-stop filters at 0.6, 0.8, and 1.0 THz with 150 GHz bandwidth

Achieved tunable rejection depths within a fixed footprint

Validated designs with FEM simulations confirming performance

Abstract

We present a genetic algorithm (GA)-based inverse design framework for synthesizing high-performance planar terahertz (THz) filters integrated with coplanar striplines (CPSs). The method efficiently explores high-dimensional design spaces to generate filter geometries matching user-defined S-parameter magnitude and phase responses, while enforcing structural connectivity for compatibility with terahertz system-on-chip (TSoC) platforms. To accelerate optimization, filter performance is evaluated using the ABCD matrix method, providing a significant computational advantage over full-wave simulations. Final validation is performed through finite element method (FEM) simulations. As a proof of concept, we design band-stop filters with center frequencies of 0.6, 0.8, and 1.0 THz, each with a 150 GHz target bandwidth, and demonstrate tunable rejection depths within a constant physical footprint. Optimization is guided by minimizing the root-mean-square error (RMSE) between simulated and target S-parameters.