Abstract: The current research on physical observable security assessment focuses on the effectiveness of physical leakage and related side-channel analysis methods. There is little research on the evaluation of the remaining workload of key cracking after side-channel attack, and a security method for key rank evaluation without prototypes has not been effectively proposed. Based on the introduction of stochastic methods and Bayesian probabilities, combined with Nicolas et al.’s research on rank evaluation method, a new security evaluation method for prototypeless side-channel analysis based on cryptographic chips is proposed. Through the correlation analysis and comparison experiment of the microcontroller （AT89C52） running AES encryption algorithm, the experimental results show that when the data size is 500, the number of traversal keys of the correlation analysis method is 20, and the number of traversal keys of the rank evaluation method based on Bayesian probability is 18, so the method proposed in this paper can quantitatively evaluate the remaining space capacity of the key, and effectively solves the key rank evaluation problem of the actual attack equipment if the prototype attack result cannot produce the probability value.

Key words: rank evaluation, security evaluation, correlation analysis, stochastic model, Bayesian probability

摘要： 针对当前物理可观测安全评估研究主要集中在物理泄露以及相关旁路分析方法的有效性方面，对进行旁路攻击后的密钥破解剩余工作量的评估研究较少，且没有有效地提出一种针对无原型的密钥秩评估安全方法。在介绍随机方法以及贝叶斯概率的基础上，结合Nicolas等人对秩评估方法的研究，提出了一种基于密码芯片的无原型旁路分析安全评估新方法。对运行AES加密算法的微控制器（AT89C52）进行相关性分析对比实验，实验结果表明，在数据规模为500时，基于相关性分析方法所需要的遍历密钥次数为20，而基于贝叶斯概率的秩评估方法遍历密钥次数为18，因此所提出的方法可以定量评估密钥剩余空间容量，有效地解决了在无原型攻击结果无法产生概率值的情况下，对实际攻击设备进行密钥秩评估的问题。

关键词: 秩评估, 安全评估, 相关性分析, 随机模型, 贝叶斯概率