电讯技术2025，Vol.65Issue(7)：1069-1077,9.DOI:10.20079/j.issn.1001-893x.241118005
基于深度学习的多密码适配轻量级密码差分分析算法
A Deep Learning-based Lightweight Cryptographic Differential Analysis Algorithm for Multi-cipher Adaptation
邓浩然 1王春琦 2杨程 1董斐斐 1陈通海1
作者信息
- 1. 国家无线电监测中心成都监测站,成都 611130
- 2. 南京航空航天大学 电子信息工程学院,南京 210016
- 折叠
摘要
Abstract
For resource-constrained Internet of Things(IoT)devices,in view of the limitation of traditional deep learning-based differential cryptanalysis methods,where the model can only be adapted to a single cipher,a deep learning-based differential cryptanalysis algorithm for multi-cipher adaptation of lightweight ciphers is proposed.By introducing two unique data input mechanisms,it can process multiple types of cipher data simultaneously,greatly enhancing its versatility.At the same time,by introducing a brand-new parameter adjustment strategy for the key recovery algorithm,the model can flexibly perform differential cryptanalysis on multiple lightweight ciphers after a single training.In addition,the proposed algorithm supports the quantitative analysis of the similarity between different ciphers and the security of alternating encryption,which is not available in traditional algorithms.Experimental results show that,compared with traditional methods,the proposed algorithm has significantly improved generalization performance and key recovery accuracy among different ciphers,with the maximum improvement reaching 2.1%.This achievement provides an efficient analysis and evaluation strategy for data security protection in future communications and is expected to play an important role in the security protection of resource-constrained IoT devices.关键词
物联网设备/安全防护/深度学习/轻量级密码/差分分析/多密码适配Key words
Internet of Things device/security protection/deep learning/lightweight cipher/differential cryptanalysis/multi-cipher adaptability分类
信息技术与安全科学引用本文复制引用
邓浩然,王春琦,杨程,董斐斐,陈通海..基于深度学习的多密码适配轻量级密码差分分析算法[J].电讯技术,2025,65(7):1069-1077,9.
