An exact reformulation-linearisation technique algorithm for solving a parameter extraction problem arising in compact thermal models

Abstract

Compact thermal models (CTMs) are used to evaluate thermal characteristics of critical electronic chip packages under several hypothetical design and usage scenarios. In order to ensure that CTMs accurately estimate thermal response, an appropriate set of model parameters must first be selected. The challenge arising in this paper is to glean a set of parameters that best describes the performance of CTM over a set of test conditions. We examine a non-convex optimisation formulation for this parameter estimation problem, and derive the first exact optimisation algorithm for the problem using a continuous version of the reformulation-linearisation technique. We provide experimental results on real data to demonstrate the applicability of our approach.

Keywords:

• branch-and-bound
• computation
• compact thermal models
• nonlinear optimisation

AMS Subject Classification :

• 90C26
• 90C57
• 90C90

Acknowledgements

The authors gratefully acknowledge the comments made by two referees, which helped to improve our paper. Dr. Smith gratefully acknowledges the support of the Air Force Office of Scientific Research under Grant #FA9550-08-1-0189.