AntennaCAT: Automated antenna design with machine learning-assisted optimization [Open Source]

As antenna design research progresses into more advanced techniques, recurrent manual editing of CAD models is one of the last, laborious obstacles for fully automated optimization. In this article we present AntennaCAT (Antenna Calculation and Autotuning Tool), which is designed to lower the barrier to entry for advanced antenna design and optimization while increasing the repeatability of the experimental process. AntennaCAT is open source software that automates the design, simulation, and optimization processes. It is compatible with multiple electromagnetic (EM) simulation software suites, including HFSS, CST, and FEKO, and includes integrated optimizers popular for multiparameter antenna design problems. Discussed are major software features for design, data collection and interpretation for machine learning (ML), internal simulation-integrated optimizers and their ML assisted hyperparameter tuning, and significance for replication studies.