Simulated Scattering Parameter Dataset for a Randomly Generated Non-Parameterized Microstrip Dual-Band Filter between 1 and 10 GHz.
Description
This dataset contains 21,000 simulated variations of a non-parameterized microstrip filter between 1 and 10 GHz. The microstrip filter is created by tapping a 16mm square patch with 2 orthogonal feedlines. Variations are introduced by perturbing the patch of the resonating structure. These variations are defined by a topology, a 32 x 32 binary image where each pixel represents a 0.5mm square section of the conductive material on the filter patch. A pixel of value 1 represents the presence, and 0 the absence, of conductive material. The topologies were generated from eight topology distributions, separated and labelled in the Topology Folder. A topology distribution is defined by a set of uniform distributions, each representing a certain shapes parameter value. The final topology is created by sampling each distribution, combining all shapes and finally discretizing the shape to form the 32 x 32 binary image. Each dataset entry pairs a topology, with the corresponding two-port S-parameters, reference impedances, and frequency points from a full-wave simulation. To obtain the electromagnetic response, the topology is used to update a filters patch and before simulating the structure using a full wave simulation software, CST Studio Suite. The MATLAB and Python folders contain an example script that can be used to import and access the data. The Topology Folder contains 8 subfolders, each representing a different topology distribution. Each subfolder contains a set of files that hold a section for the dataset, each file contains a a 3 fields, namely imageName, topology and solution. Each subfolder also contains an Images folder. This folder contains one image for each topology, where the images name correlates to the imageName field in the datafile, these can be helpful to a user to identify a specific topology.
Files
Steps to reproduce
The attached Github repository contains the updated scripts, guide and CST model used to generate, process and access the dataset. The guide can be used to recreate or expand the existing dataset.
Institutions
- Stellenbosch UniversityWestern Cape, Stellenbosch