A static bidirectional learning technique to accelerate test pattern generation

Abstract

Learning is a very effective technique to speed up the test pattern generation process. In this paper, we propose a static bidirectional learning technique that significantly increases the number of learned implications. The proposed bidirectional learning is integrated into PODEM. Experiment results show that, compared to previous static learning, in average the number of learned implications increases by 10%, and the CPU times are reduced by 24%.