New crossover operators for GA-based synchronous sequential circuit testing

Abstract

Testing is an essential part of any VLSI manufacturing system as it is necessary to separate bad circuits from the good ones. The testing process detects the physical defects produced during fabrication. Test generation for sequential circuits is a search problem over large vector space, proportional to the number of inputs and number of states. Genetic Algorithm (GA) has been effectively used for Automatic Test Pattern Generation. In this paper, two new crossover operators, the sequence based and the weight-based, are proposed. The proposed operators significantly improve the performance of a GA-based test pattern generator for stuck-at-faults in synchronous sequential circuits. The effectiveness of the approach is demonstrated on the ISCAS89 sequential benchmark circuits. Experimental results on the ISCAS89 sequential benchmark circuits demonstrate that the weight-based crossover operator is effective for GA-based test generation. The weight-based crossover provides high fault coverage and compact test vectors for most of the benchmark circuits.