EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Netlist Recognition Systems Using Graph Neural Network for Analog Integrated Circuit Design and their Performance Evaluation

Arif Abdul Mannan, Koichi Tanno

International Journal on Advanced Science, Engineering and Information Technology · 25.06.2026 · 16 (3) · 957-966

DOI: 10.18517/ijaseit.16.3.21106 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal on Advanced Science, Engineering and Information Technology, online erschienen am 25.06.2026. Verfasst von Arif Abdul Mannan und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

In recent years, vast researchers have developed artificial intelligence (AI) to assist the analog integrated circuit (IC) designer, especially to increase the final IC performance after manufacture. However, the design process itself still requires a high level of expertise and experience, either from a topological point of view or from parameter optimization. In this paper, a netlist recognition system is proposed as a part of the bigger Complementary Metal-Oxide-Semiconductor (CMOS) analog circuit design warning system to support the designer in topological analog circuit design checking, utilizing the combination between the proposed novel netlist to graph abstraction technique, to change the text-based circuit netlist into its graph representation, and some already known graph neural network (GNN) architecture for the recognition system. This paper focuses on the netlist-to-graph abstraction technique, specifically designed for analog circuit design, which eliminates the limitations of the previous netlist-to-graph method. From the simulation results, the novel netlist-to-graph abstraction technique can achieve 100% recognition accuracy with a graph convolutional network, a graph isomorphism network, and GraphSAGE, and 99.8% accuracy with a graph attention network. Compared to previous methods, the proposed technique achieves higher accuracy and greater recognition stability on analog-specific datasets. The GNN performance, including the final dataset size, the detailed distribution of recognition errors, the recognition accuracy, and the GNN's single-epoch recognition time, is presented to provide a simple picture of its application performance in the CMOS analog circuit warning system.

Zitieren

APA

Mannan, A. A., & Tanno, K. (2026). Netlist Recognition Systems Using Graph Neural Network for Analog Integrated Circuit Design and their Performance Evaluation. International Journal on Advanced Science, Engineering and Information Technology, 16(3), 957-966. https://doi.org/10.18517/ijaseit.16.3.21106

Vancouver

Mannan AA, Tanno K. Netlist Recognition Systems Using Graph Neural Network for Analog Integrated Circuit Design and their Performance Evaluation. International Journal on Advanced Science, Engineering and Information Technology. 2026;16(3):957-966. doi:10.18517/ijaseit.16.3.21106

BibTeX
@article{Mannan2026,
  author = {Mannan, Arif Abdul and Tanno, Koichi},
  title = {Netlist Recognition Systems Using Graph Neural Network for Analog Integrated Circuit Design and their Performance Evaluation},
  journal = {International Journal on Advanced Science, Engineering and Information Technology},
  year = {2026},
  volume = {16},
  number = {3},
  pages = {957--966},
  doi = {10.18517/ijaseit.16.3.21106}
}
Kostenlos starten →