具有优良可靠的侧信道安全性是高安全等级密码芯片必须满足的一项共性技术要求。近期，“类幽灵”和“类熔断”等芯片级重大实现漏洞先后披露，给学术界和工业界造成了巨大震动，再次彰显出侧信道漏洞分析与风险评估对保障芯片安全可控的极端重要性。从理论上讲，基于深度学习的侧信道分析方法具有为密码芯片安全分析测评实践提供新颖解决方案和精选先进技术的巨大潜能。因此，项目拟以能量分析攻击和/或电磁分析攻击为主要分析手段，开展基于深度学习的侧信道分析方法研究，主要研究内容包括基于深度学习的信息泄漏预处理方法、基于深度学习的新颖区分器构造方法以及基于深度学习的侧信道分析软件工具包研制等。项目研究旨在为审视密码芯片侧信道安全风险威胁与潜在漏洞提供高级方法支持，为建立科学先进的密码芯片侧信道安全性评测技术体系提供核心工具支撑。

It is a common technical requirement to have well-designed or even formally verified side-channel security for higher security level cryptographic chips. Recently, the discovery of Meltdown-like and Spectre-style implementation vulnerabilities in modern processors has led to a considerable shock in both academia and industry, which once again highlights the importance of seriously examining the side-channel security of integrated circuits. Therefore, vulnerability analysis and security evaluation for cryptographic chips against side channel attacks are extremely crucial to ensure the secure, reliable and controllable cryptosystems. In theory, deep learning based side-channel analysis approaches have great potential to provide novel solutions and advanced technologies in practical evaluation of cryptographic chips..Therefore, the proposed project aims to investigate deep learning based side-channel analysis approaches on cryptographic modules by using the methods of power analysis and/or electromagnetic analysis. The proposed research topics include: to develop deep learning based information leakages preprocessing methods, to propose deep learning based novel distinguishes construction methods, and to develop the deep learning based side-channel analysis software toolkit..The main research goal of this project is to explore novel approaches and cutting-edge techniques to support the evaluation of side-channel security of cryptographic chips, and to promote the establishment of advanced side-channel security evaluation technology system.