DKEMA: GPU-based and dynamic key-dependent efficient message authentication algorithm

Abstract

Recently, benefiting from the advancement in the Graphic Processing Unit (GPU) technology, there is an increased interest in implementing and designing new efficient cryptographic schemes. Existing cryptographic algorithms, especially the Message Authentication Algorithms (MAAs) such as Hash Message Authentication Code (HMAC) and Ciphered Message Authentication Code (CMAC), are not designed to benefit from the GPU characteristics, which results in degraded performance of their GPU implementations. This gives rise to a trade-off between the design concept and the performance level. In this paper, a new MAA, called ’DKEMA’, is proposed to better suit the GPU functionality. This scheme is based on the dynamic key-dependent scheme with one round of substitution and diffusion operations. The experimental results show that the proposed solution is highly effective on Tesla V100 and A100 GPUs, and the throughput is, respectively, more than 400GB/s and 500GB/s. Therefore, DKEMA can be considered as a promising MAA candidate for GPU implementation, achieving the desired cryptographic properties such as high randomness, collision tolerance in addition to message and key avalanche effect. The experimental results show that the proposed solution, based on the dynamic key approach, is immune towards well-known authentication and cryptanalysis attacks. In addition, DKEMA, consisting of one round compression function, presents an enhancement in terms of performance compared to existing algorithms (e.g. AES and SHA). The code is available in GitHub: https://github.com/rcouturier/GPU-DKEMA.git. © 2022, The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature.