论文部分内容阅读
基于数据挖掘思想,提出一种链状结构模板的规律性提取算法,解决集成电路规律性提取算法复杂度过高的问题.通过对边权值进行编码,将复杂子电路的同构搜索转化为边权值序列的匹配问题.模板扩展过程利用剪枝策略删除非频繁子电路,提高了规律性提取效率.将模板的产生与子电路的同构搜索过程合并,简化规律性提取流程.解决大规模集成电路中规则性提取复杂度过高的问题.结果表明,算法比SPOG与TREE算法更能充分提取电路的规律性,得到较好的电路覆盖.
Based on the idea of data mining, this paper proposes a regularity extraction algorithm of chain structure template, which solves the problem of high complexity of ICR algorithm.Based on data mining, the isomorphism search of complex sub-circuits is transformed into Edge weight sequence matching problem.Extensive pruning process uses pruning strategy to delete non-frequent sub-circuits and improve the efficiency of regular extraction.The template generation and sub-circuit isomorphism search process is combined to simplify the regular extraction process. The problem of excessive complexity of rule extraction in large-scale integrated circuits is found.The results show that the algorithm can extract the regularity of the circuit more fully than the SPOG and TREE algorithms, and get a better circuit coverage.