A Novel Fuzzing Mutation Scheduling Method based on Evolutionary Strategy

Aiming at the problem of optimizing mutation-based fuzzing algorithms, this paper applies an evolutionary strategy specifically tailored to the havoc stage. By refining the mutation scheduling within the currently most popular fuzzing tool AFL, we seek to approximate the ideal probability distribution of selecting mutation operators. Therefore, it could avoid the waste of fuzzing iterations on suboptimal mutation operators. Experimental evaluations conducted on three real-world test programs reveal a pronounced improvement in crash detection, highlighting the potential of this strategy to enhance fuzzing efficiency and effectiveness. At the same time, in order to better utilize testing tools and visually display test results, a usage platform has been built successfully.