Welcome to the resource topic for 2021/362
Title:
Cryptanalysis of Round-Reduced SIMON32 Based on Deep Learning
Authors: Zezhou Hou, Jiongjiong Ren, Shaozhen Chen
Abstract:Deep learning has played an important role in many fields. It shows significant potential to cryptanalysis. Differential cryptanalysis is an important method in the field of block cipher cryptanalysis. The key point of differential cryptanalysis is to find a differential distinguisher with longer rounds or higher probability. Firstly, we describe how to construct the ciphertext pairs required for differential cryptanalysis based on deep learning. Based on this, we train 9-round and 8-round differential distinguisher of SIMON32 based on deep residual neural networks. Secondly, we explore the impact of the input difference patterns on the accuracy of the distinguisher based on deep learning. For the input difference with Hamming weight of 1, the accuracy of 9-round distinguisher is different between the first 16 bits and the last 16 bits for non-zero bit positions. This is mainly caused by that its nonlinear operation is mainly concentrated in the first 16 bits. We also find that the accuracy of the distinguisher is different even if the input differences come from the differential characteristics with the same probability. Finally, we construct a last subkey recovery attack on 11-Round SIMON32 with practical data and time complexities. Our attack only uses about 29 chosen plaintexts and only needs about 45s for an attack with a success rate of over 90% using our workstation, which does not exceed 2^18:5 11-round encryption. At the same time, we extend the neural 9-round distinguisher to a 11-round distinguisher based on SAT, and propose a last subkey recovery attack on 13-Round SIMON32 using 2^12:5 chosen plaintexts with a success rate of over 90%. Compared with traditional approach, the complexity of the method based on deep learning is lower, both in time complexity and data complexity.
ePrint: https://eprint.iacr.org/2021/362
See all topics related to this paper.
Feel free to post resources that are related to this paper below.
Example resources include: implementations, explanation materials, talks, slides, links to previous discussions on other websites.
For more information, see the rules for Resource Topics .