Comprehensive coverage measurement should guide an effective testbench generation approach. Today, feedback from coverage to test generation often requires manual work; it is desirable to implement a framework that automates this feedback process. The authors propose a genetic-algorithm-based evolution framework for testbench generation. It enables small test programs to evolve and effectively capture target corner cases based on the feedback from coverage measurement.
Citation:
Fulvio Corno, Ernesto S?nchez, Matteo Sonza Reorda, Giovanni Squillero, "Automatic Test Program Generation: A Case Study," IEEE Design and Test of Computers, vol. 21, no. 2, pp. 102-109, Mar./Apr. 2004, doi:10.1109/MDT.2004.1277902