Efficient Random Generating of Bent Functions Using Hybrid CPU-GPU Platform

Bent functions are a significant group of Boolean functions since they have applications in various cryptographic algorithms. General form of bent functions is still unknown, and it is very demanding to generate them when they have many variables. That's why it exists various interesting methods for generating these functions from certain classes of bent functions. However, the functions thus obtained are deterministic. Another approach for generating has been the use of the random discovering of bent functions. In this way, non-deterministic bent functions are obtained. Bent function discovering has been shown to be most efficient when it is performed in the Reed-Muller domain and in parallel. Therefore, this paper proposes method for random discovering of bent functions in the Reed-Muller domain using both multicore CPUs (central processing units) and GPUs (graphics processing units) for performing the computations in parallel. The experimental comparison of the proposed method and previously developed method using only GPU, confirms the efficiency of the method.