DCIS01

Using Tabu Search Method for Optimizing the Cost of Hybrid BIST

Raimund Ubar
 
Helena Kruus
Gert Jervan Author homepage
 
Zebo Peng Author homepage

16th Conference on Design of Circuits and Integrated Systems (DCIS 2001), Porto, Portugal, November 20-23, 2001, pp. 445-450

ABSTRACT
This paper deals with a hybrid BIST solution for testing systems-on-chip, which combines pseudo-random test patterns with stored precomputed deterministic test patterns. A method is proposed for finding the optimal balance between pseudorandom and stored test patterns to perform core test with minimum cost of time and memory, and without losing test quality. As a generalization of local optimization, Tabu search method is used for finding the optimal balance. Unlike local search which stops when no improved new solution is found in the current neighborhood, tabu search continues the search from the best solution in the neighborhood even if it is worse than the current solution. To speed up the optimization procedure, a fast method for predicting the location of optimum solution is also used. Experimental results on benchmark circuits have proved the efficiency of the proposed approach for BIST optimization.


Related files:
DCIS01.pdfAdobe Acrobat portable document
DCIS01.pspostscript document


[UKJP01] Raimund Ubar, Helena Kruus, Gert Jervan, Zebo Peng, "Using Tabu Search Method for Optimizing the Cost of Hybrid BIST", 16th Conference on Design of Circuits and Integrated Systems (DCIS 2001), Porto, Portugal, November 20-23, 2001, pp. 445-450
( ! ) perl script by Giovanni Squillero with modifications from Gert Jervan   (v3.1, p5.2, September-2002-)
Last modified on Monday December 04, 2006 by Gert Jervan