|
| Abstract: |
| As process technology development, model order reduction (MOR) has been regarded as a useful tool in analysis of on-chip interconnects. We propose a weighted self-adaptive threshold wavelet interpolation MOR method on account of Krylov subspace techniques. The interpolation points are selected by Haar wavelet using weighted self-adaptive threshold methods dynamically. Through the analyses of different types of circuits in very large scale integration (VLSI), the results show that the method proposed in this paper can be more accurate and efficient than Krylov subspace method of multi-shift expansion point using Haar wavelet that are no weighted self-adaptive threshold application in interest frequency range, and more accurate than Krylov subspace method of multi-shift expansion point based on the uniform interpolation point. |
| Key words: interconnect model order reduction Haar wavelet transform weighted threshold multi shift Arnoldi circuit synthesis |
| DOI：10.11916/j.issn.1005-9113.16119 |
| Clc Number: |
| Fund: |
|
| Descriptions in Chinese: |
| 基于加权自适应阈值小波的互连模型降阶插值点选择方法 王新胜，喻明艳 （哈尔滨工业大学 航天学院，哈尔滨 150001） 创新点说明：提出一种加权自适应阈值Haar小波插值点选择的Krylov子空间模型降阶技术，其能动态的选择插值点的阈值，从而有效保证集成电路互连模型降阶的精度和速度。 研究目的： 有效解决大规模集成电路互连Krylov子空间模型降阶的精度与速度折中问题。 研究方法： 采用加权自适应阈值Haar小波插值点选择Krylov子空间模型降阶技术。 结果： 通过仿真分析大规模集成电路中电源和时钟树网络寄生电路系统，结果表明加权自适应阈值Haar小波插值点选择的Krylov子空间模型降阶技术相比传统的多偏移插值点选择的Krylov子空间模型降阶技术能有效提高降阶精度和降阶速度。 结论： 本文提出一种加权自适应阈值Haar小波插值点选择的Krylov子空间模型降阶技术，其能有效保证集成电路互连模型降阶的精度和速度 关键词：互连模型降阶；Haar小波变换；加权阈值；多偏移Arnoldi方法；电路综合 |
