A high-efficiency modeling method for analog integrated circuits

Chip Pub Date : 2025-03-07 DOI:10.1016/j.chip.2025.100135
Dongdong Chen , Yunqi Yang , Xianglong Wang , Di Li , Guoqing Xin , Yintang Yang
{"title":"A high-efficiency modeling method for analog integrated circuits","authors":"Dongdong Chen ,&nbsp;Yunqi Yang ,&nbsp;Xianglong Wang ,&nbsp;Di Li ,&nbsp;Guoqing Xin ,&nbsp;Yintang Yang","doi":"10.1016/j.chip.2025.100135","DOIUrl":null,"url":null,"abstract":"<div><div>Integrated circuits (ICs) are the foundation of information technology development. The optimal design scheme of an analog IC is determined by iteratively running the simulation software and comparing the performance metrics. However, the simulation software of an analog IC is time-consuming, which leads to the low design efficiency. Due to the nonideal factors in analog ICs, the nonlinear relationship between design parameters and performance metrics cannot be well described by the deduced approximation equations. Inspired by the image and semantic recognition, a universal high-efficiency modeling method for analog ICs based on convolutional neural network (CNN) was proposed in the current work, named as CNN-IC. The sparse topology mapping method was proposed to map the design parameters into a sparse matrix, which includes the spatial and transistor characteristics of analog IC. The CNN model with three convolutional kernels was constructed to extract “transistor-circuit module-integrate circuit” features level by level, which can replace the simulation software to effectively improve the training efficiency and accuracy. Two typical analog ICs were selected to verify the effectiveness of the CNN-IC model. The results show that the accuracy of the CNN-IC model could reach over 99% and that its convergence rate was the fastest compared with the machine learning models in the state of the art.</div></div>","PeriodicalId":100244,"journal":{"name":"Chip","volume":"4 3","pages":"Article 100135"},"PeriodicalIF":0.0000,"publicationDate":"2025-03-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Chip","FirstCategoryId":"1085","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S2709472325000097","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0

Abstract

Integrated circuits (ICs) are the foundation of information technology development. The optimal design scheme of an analog IC is determined by iteratively running the simulation software and comparing the performance metrics. However, the simulation software of an analog IC is time-consuming, which leads to the low design efficiency. Due to the nonideal factors in analog ICs, the nonlinear relationship between design parameters and performance metrics cannot be well described by the deduced approximation equations. Inspired by the image and semantic recognition, a universal high-efficiency modeling method for analog ICs based on convolutional neural network (CNN) was proposed in the current work, named as CNN-IC. The sparse topology mapping method was proposed to map the design parameters into a sparse matrix, which includes the spatial and transistor characteristics of analog IC. The CNN model with three convolutional kernels was constructed to extract “transistor-circuit module-integrate circuit” features level by level, which can replace the simulation software to effectively improve the training efficiency and accuracy. Two typical analog ICs were selected to verify the effectiveness of the CNN-IC model. The results show that the accuracy of the CNN-IC model could reach over 99% and that its convergence rate was the fastest compared with the machine learning models in the state of the art.
模拟集成电路的一种高效建模方法
集成电路是信息技术发展的基础。通过对仿真软件的迭代运行和性能指标的比较,确定模拟集成电路的最佳设计方案。然而,模拟集成电路的仿真软件耗时长,导致设计效率低。由于模拟集成电路中存在非理想因素,推导出的近似方程不能很好地描述设计参数与性能指标之间的非线性关系。受图像和语义识别的启发,本文提出了一种基于卷积神经网络(CNN)的模拟集成电路通用高效建模方法,命名为CNN- ic。提出稀疏拓扑映射方法,将设计参数映射到包含模拟IC空间特性和晶体管特性的稀疏矩阵中,构建具有3个卷积核的CNN模型,逐级提取“晶体管-电路模块-集成电路”特征,可替代仿真软件,有效提高训练效率和精度。选择了两个典型的模拟ic来验证CNN-IC模型的有效性。结果表明,CNN-IC模型的准确率可以达到99%以上,其收敛速度是目前最先进的机器学习模型中最快的。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
CiteScore
2.80
自引率
0.00%
发文量
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:604180095
Book学术官方微信