Skip to main navigation Skip to search Skip to main content

Time domain identification of passive multiport RC networks with convex optimization: An application to thermal impedance macromodeling

  • L. De Tommasi
  • , A. Magnani
  • , V. D'Alessandro
  • , M. De Magistris

Research output: Chapter in Book/Report/Conference proceedingsConference proceedingpeer-review

Abstract

In this paper a novel time-domain identification procedure for passive RC multiport networks is introduced, based on convex optimization and allowing the straightforward enforcement of basic physical constraints, specifically tailored to thermal impedances. The identification is performed in two steps. First, the time constants are identified for each element of the thermal impedance matrix by means of standard nonlinear optimization techniques, which allows ensuring that all corresponding network poles are real. Subsequently, a matrix form for the response is assembled with the evaluated common set of time constants, and residue matrices are computed through a convex optimization algorithm, minimizing the approximation error while enforcing the passivity constraints.

Original languageEnglish
Title of host publication2014 18th IEEE Workshop on Signal and Power Integrity, SPI 2014 - Proceedings
PublisherIEEE Computer Society
ISBN (Print)9781479935994
DOIs
Publication statusPublished - 2014
Externally publishedYes
Event18th IEEE Workshop on Signal and Power Integrity, SPI 2014 - Ghent, Belgium
Duration: 11 May 201414 May 2014

Publication series

Name2014 18th IEEE Workshop on Signal and Power Integrity, SPI 2014 - Proceedings

Conference

Conference18th IEEE Workshop on Signal and Power Integrity, SPI 2014
Country/TerritoryBelgium
CityGhent
Period11/05/1414/05/14

Keywords

  • Convex optimization
  • Identification techniques
  • RC network
  • Thermal impedance

Fingerprint

Dive into the research topics of 'Time domain identification of passive multiport RC networks with convex optimization: An application to thermal impedance macromodeling'. Together they form a unique fingerprint.

Cite this