Название: Efficient Analog Integrated Circuit Sizing with GenAIАвтор: Pedro H.M. Eid, Filipe P. Azevedo, Nuno C.C. Lourenço, Ricardo M.F. Martins
Издательство: Springer
Год: 2025
Страниц: 92
Язык: английский
Формат: pdf (true), epub
Размер: 21.8 MB
This book focuses on the automation of analog integrated circuit design, particularly the sizing process. It introduces an innovative approach leveraging Generative Artificial Intelligence, specifically denoising diffusion probabilistic models (DDPM). The proposed methodology provides a robust solution for generating circuit designs that meet specific performance constraints, offering a significant improvement over conventional techniques. By integrating advanced Machine Learning models into the design workflow, the book showcases a transformative way to streamline the process while maintaining accuracy and reliability.
In recent years the evolution of the electronics industry, and the increase of the demand for consumer-graded products, like smartphones, and computers, has resulted in a reduction in the size and an increase in complexity of IC. While these advancements have propelled the field forward, they have concurrently led the way in a surge of intricacies associated with the design of IC. The challenge of the design can be significant when there is a need for analog parts on the IC, like in a Mixed-Signal System-On-Chip. While the digital aspect has benefited from streamlined design flows through Computer-Aided Design, the field of Electronic Design Automation for analog IC has faced challenges in keeping pace with this complexity. Presently, the conventional design flow for analog circuits requires human intervention due to the absence of automation tools capable of accommodating non-linear components, bias requirements, and real-world effects such as stray impedance. This gap in Computer-Aided Design capabilities has transformed the design of analog circuits into the bottleneck of the IC design. Despite the smaller physical footprint occupied by analog circuits within IC compared to their digital counterparts, the absence of robust automation solutions renders the formers design notably more challenging.
This book contributes to the field of electronic design automation. Its primary focus lies in automating the design of analog integrated circuits, with a particular emphasis on the sizing task of the process. It proposes to leverage ANN, particularly utilizing diffusion models, to enhance and streamline the automation process. Researchers have explored various automation methods, including meta-heuristics and optimization-based approaches, to address this challenge. However, each method presents distinct drawbacks and, at times, yields inefficient results. While studies have made some attempts using ANN, they commonly face the hurdle of the ill-posed nature problem exacerbated by the scarcity of databases for training the models. This work introduces a novel approach based on state-of-the-art Generative Artificial Intelligence to automate the design process, by leveraging diffusion models to enhance the existing ANNs-based framework and address the limitations of previous methodologies. Specifically, Denoising Diffusion Probabilistic Models (DDPM) to tackle the inverse problem of circuit sizing. DDPM employ a noising and denoising architecture, where they learn to reconstruct input distributions by progressively adding and removing noise, starting from Gaussian white noise. Once trained, the DDPM can generate new data from pure noise. Our approach uses this capability to generate new circuit sizing solutions while satisfying performance constraints.
Скачать Efficient Analog Integrated Circuit Sizing with GenAI