A methodology to generate evenly distributed input stimuli by clustering of variable domain

Constrained Random Verification (CRV) is becoming the mainstream methodology for the functional verification of complex System on Chip (SoC) designs. In order to achieve verification closure, CRV tools have to produce a large number of solutions, evenly distributed, in the search space. To attain this requirement, we propose a technique which analyzes the solution space by using consistency algorithm and splits the variable´s domain into clusters. The proposed technique helps to generate input stimuli which are evenly distributed in search space. The proposed technique has been validated through experimental results. Experimental results show that the proposed methodology guarantees evenly distributed stimuli and improves by about 15% the veriication coverage.