Research On Template Attack Method Based On Secondary Classification

Template attack is a strong attack method based on the analysis of power consumption information,which has become one of the most popular research directions in the field of side channel research.And the attack method is a great threat to cryptographic equipment.With the upgrading and improvement of cryptographic algorithms,the difference between the leaked data collected through side-channel devices is getting smaller,and the success rate of traditional template attack methods is reduced accordingly.Some classification algorithms in machine learning and some steps of template attack play a similar role.Therefore,researchers apply these algorithms into the template attack implementation to improve the attack performance.In the traditional machine learning based template attack methods,whether the Hamming weight or the intermediate value is as the label,the classification result will lead to complicated subsequent calculations and reduce the accuracy of the model.In view of the issue,the main research contents of this paper are as follows:(1)Because of the small differences in collecting power consumption information,the success of traditional template attack is lower.In this case,the paper designs and implements a template attack system based on Euclidean distance algorithm.The Euclidean distance algorithm can more accurately calculate the distance between two points,and it can improve the accuracy of template construction and the success rate of matching in the feature selection and template matching stages.The experimental results show that the attack success rate of the system has been improved.(2)In decision tree-based template attack methods,the accuracy of the model will decrease when the number of energy traces is limited.For solving this problem,this paper proposes a template attack method based on the secondary classification algorithm,and uses the decision tree as the classification algorithm to realize the template attack.In the first classification,the energy trace is labeled with the Hamming weight,and the data is classified to reduce the key search range.In the second classification,the intermediate value of the energy trace is used to classify the label,and finally the key can be recovered.It is validated show that this method has a higher attack success rate than the traditional decision tree-based template attack methods.(3)Support vector machine can effectively avoid the traditional data induction and deduction process,simplify the classification problem to a large extent,and has good robustness.The template attack method based on secondary classification uses support vector machine(SVM)as the classification algorithm for template attack.The experimental results show that this method has a higher attack success rate than the existing SVM-based template attack methods.