基于遗传算法的覆盖率驱动测试产生器  

Coverage directed test generation based on genetic algorithm

作　　者：王树朋[1] 黄凯[1] 严晓浪[2] 

机构地区：[1]浙江大学信息与电子工程学系,浙江杭州310027 [2]浙江大学超大规模集成电路研究所,浙江杭州310027

出　　处：《浙江大学学报（工学版）》2016年第3期580-588,共9页Journal of Zhejiang University：Engineering Science

基　　金：国家自然科学基金资助项目(61100074);核高基国家科技重大专项资助项目(2012ZX01039-004);中央高校基础研究基金资助项目(2013QNA5008)

摘　　要：为了更好地建立覆盖率和测试产生器之间的联系,产生高质量的测试,提出基于遗传算法的覆盖率驱动测试产生器.该测试产生器利用一种简单、准确的测试编码方法对测试进行编码,并利用基于功能覆盖率的适应度函数评估测试的优劣.通过遗传算法(GA)建立覆盖率与测试产生器之间的联系,分析覆盖率和测试之间的关系,根据分析结果改变测试产生器的约束和限制,驱动测试产生器生成新一代的测试,新一代的测试可以覆盖到上一代的测试无法覆盖的功能点.实验结果表明:在2个高性能的32位多核处理器的验证环境中,该测试产生器可以明显减少仿真时间,提高验证效率.Coverage directed test generation based on genetic algorithm（GA）was proposed to close the loop between coverage analysis and test generation and produce the tests of good quality.A simple and accurate test encoding method was proposed.A fitness function based on functional coverage was used to evaluate the quality of tests.GA was used to close the loop between coverage analysis and test generation.The coverage results were evaluated and the constraints for test generation were modified to direct the test generation to produce the new tests,which can cover the functions that the old tests can＇t cover.The experiments were conducted based on the simulation environment for verifying two high-performance 32-bit multi-core processors.Results show that the proposed method can significantly reduce simulation time and improve verification efficiency.

关 键 词：覆盖率 测试产生器 遗传算法(GA) 覆盖率驱动测试产生器 适应度函数 

分 类 号：TN47[电子电信—微电子学与固体电子学]