传统的状态覆盖方法对电路的数据单元测试不足,而随机测试方法又具有盲目性.在综合2种方法的基础上,给出一种以状态与状态转换覆盖率为评估、以遗传筛选为工具对生成的测试向量进行择优选择的方法.为了指导测试生成,给出了动态状态转换与静态状态转换概念.同时,基于该方法给出一个测试生成工具GRTT.最后,将文中方法实验于ITC99-benchmark电路,并将实验结果与测试生成系统X-Pulling的结果进行比较.
The merit and shortcoming of traditional states covering method and genetic method are analyzed in this article. A new genetic selecting approach is presented to overcome the shortcoming of these two methods. First, it can be implemented at RT-level. Second, it uses state coverage as fitness function, which is useful to test the control-part of the circuit. Third, it can test the control part and data part of circuit at the same time. The concept about dynamic state transfer and static state transfer are also brought up in this paper to direct test pattern generation. Based on this approach, an ATPG toot named GRTT is developed. Experimental results on ITC99-benchmarks show that GRTT can get excellent results not only in coverage but also in run-time. In comparison with X-Pulling, an experimental RT-level ATPG system, GRTT runs faster.