Enhancing SAT-Based Test Pattern Generation

刘歆,熊有伦
2005-01-01
Journal of Electronic Science and Technology
Abstract:This paper presents modeling tools based on Boolean satisfiability (SAT) to solve problems of test generation for combinational circuits. It exploits an added layer to maintain circuit-related information and value justification relations to a generic SAT algorithm. It dovetails binary decision graphs (BDD) and SAT techniques to improve the efficiency of automatic test pattern generation (ATPG). More specifically, it first exploits inexpensive reconvergent fanout analysis of circuit to gather information on the local signal correlation by using BDD learning, then uses the above learned information to restrict and focus the overall search space of SAT-based ATPG. Its learning technique is effective and lightweight. The experimental results demonstrate the effectiveness of the approach.
What problem does this paper attempt to address?