Digital circuit crosstalk delay faults test generation based on neural networks

Crosstalk faults arise easily between adjacent signal lines in integrated circuit. The faults make circuit´s output signal delay. This paper proposed an effective test generation algorithm for the crosstalk fault. Kernighan-Lin-Fiduccia-Mattheyses(KLFM) method is used to cut the circuit into two parts including the left and right parts. For the right part, the fault can be propagated to the output using neural networks method, the test patterns obtained by the method become a constraint condition; for the left part, a crosstalk fault equation can be obtained. The patterns that satisfy the constraint condition and make the delay equation have maximal value are the test generation patterns for the crosstalk fault. The experimental results demonstrate that the algorithm´s fault coverage can reach 96% and the algorithm´s test generation time is less than that of other references.