Design and security analysis of two robust keyed hash functions based on chaotic neural networks

In this paper, we designed, implemented, and analyzed the performance, in terms of security and speed, of two proposed keyed Chaotic Neural Network (CNN) hash functions based on Merkle–D $$\mathring{\rm a}$$ a ˚ mgard (MD) construction with three output schemes: CNN-Matyas–Meyer–Oseas, Modified CNN-Matyas–Meyer–Oseas, and CNN-Miyaguchi–Preneel. The first hash function’s structure is composed of two-layer chaotic neural network while the structure of the second hash function is formed of one-layer chaotic neural network followed by non-linear layer functions. The obtained results of several statistical tests and cryptanalytic analysis highlight the robustness of the proposed keyed CNN hash functions, which is fundamentally due to the strong non-linearity of both the chaotic systems and the neural networks. The comparison of the performance analysis with some chaos-based hash functions of the literature and with standard hash functions make the proposed hash functions suitable for data integrity, message authentication, and digital signature applications.