Scholarlink

<p>This paper presents the security evaluation of Randomized Butterfly architecture of fast Fourier transform for Key (RBFK) cipher using Artificial Neural Networks (ANN). RBFK is a very recent lightweight block cipher primarily proposed for IoT devices and has been widely popular for its simple and lightweight architecture. In this paper, we propose two neural network-based attacks on the RBFK cipher. The first attack model was designed using an ANN, and the second was designed using a Gaussian Basis Function. We named the network the Gaussian Basis Neural Network (GbNN). The experimental results showed that GbNN performed better at predicting the plaintext from a given ciphertext. Both networks correctly identified the majority of bits in a random plaintext of the corresponding ciphertext. Finally, the experimental results indicate that RBFK raises major security concerns in the proposed neural cryptanalysis environment.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Neural Cryptanalysis of the RBFK Cipher: A Feature-Lifted Learning Approach

  • 期刊论文

  • 发表日期：

    2026年7月1日

查看全文
  • Subinoy Sikdar,
  • Malay Kule

摘要

This paper presents the security evaluation of Randomized Butterfly architecture of fast Fourier transform for Key (RBFK) cipher using Artificial Neural Networks (ANN). RBFK is a very recent lightweight block cipher primarily proposed for IoT devices and has been widely popular for its simple and lightweight architecture. In this paper, we propose two neural network-based attacks on the RBFK cipher. The first attack model was designed using an ANN, and the second was designed using a Gaussian Basis Function. We named the network the Gaussian Basis Neural Network (GbNN). The experimental results showed that GbNN performed better at predicting the plaintext from a given ciphertext. Both networks correctly identified the majority of bits in a random plaintext of the corresponding ciphertext. Finally, the experimental results indicate that RBFK raises major security concerns in the proposed neural cryptanalysis environment.