On area minimization of complex combinational circuits using cartesian genetic programming

The paper deals with the evolutionary post synthesis optimization of complex combinational circuits with the aim of reducing the area on a chip as much as possible. In order to optimize complex circuits, Cartesian Genetic Programming (CGP) is employed where the fitness function is based on a formal equivalence checking algorithm rather than evaluating all possible input assignments. The standard selection strategy of CGP is modified to be more explorative and so agile in very rugged fitness landscapes. It was shown on the LGSynth93 benchmark circuits that the modified selection strategy leads to more compact circuits in roughly 50% cases. The average area improvement is 24% with respect to the results of conventional synthesis. Delay of optimized circuits was also analyzed.