Design and application of fault injection attack platform with a fault probability analysis function
[Objective]With the increased implementation of cryptographic algorithms in cryptographic devices as hardware circuits,low-cost physical attacks against these devices have gained importance.Among them,fault attacks have low cost and high efficiency and have gradually become one of the most effective means of current integrated circuit security attacks.However,in real-world high-frequency and high-noise environments,attackers typically require extensive data support and high computing costs,thus reducing the attack efficiency.[Methods]Given the large data requirements,high computational complexity,and poor portability,this study considers the AES-128 algorithm implemented based on microcontrol units as the research object.It combines the data dependence of fault probability and researches the attack methods in fault attacks.[Results&Conclusions]The main research work is as follows:(1)To address the significant demand for template attack data and the low portability of the method,a fault template attack based on fault probability is proposed.Fault probability traces are established,points of interest are selected,the hamming weight of the matching template is inversely calculated,and finally,the intersection is determined for key recovery.The method applies to all substitution-permutation network cipher algorithms,thereby improving the universality of the attack method.Experiments show that this method shows superior performance over traditional template attack methods and related improved methods,such as the key advantage template attack against the AES-128 algorithm and amplified template attacks using the hamming weight model.(2)A mutual information analysis based on fault probability is proposed to address the significant demand for mutual information analysis data and low attack efficiency.Calculating the mutual information between the measured and simulated leakage values to recover the key requires detailed knowledge of the algorithm and a few specific assumptions consistent with the situation practically faced by the attackers.The experimental results demonstrate the superior performance of this method over traditional mutual information analysis methods and related improved methods such as neural mutual information analysis methods.(3)Correlation analysis requires enormous data and has high computational complexity;thus,a fault probability correlation analysis based on secondary filtering is proposed.Random plaintext and sampling points are selected using the difference summary,which reduces the number of calculation iterations.The experimental results show that the method outperforms traditional correlation analysis,multiple sieve correlation power analysis,and related improved methods such as block-oriented correlation power analysis.
fault injection attackfault probabilitytemplate attackmutual information analysiscorrelation analysis
NETL
NSTL
万方数据
