Generating highly nonlinear Boolean functions using a genetic algorithm

In this paper a few algorithms are presented which assist with finding Boolean functions with good cryptographic properties, especially with high nonlinearity. First, a basic hill-climbing algorithm is described which improve the nonlinearity of a Boolean function. Then this algorithm is modified to incorporate a genetic algorithm. It is shown that these new search techniques are extremely powerful when compared to traditional random search techniques. Experimental results successfully prove this statement.