Side Channel Analysis And Implementation Based On Block Cipher Algorithm

Side Channel Analysis(SCA)is an emerging cryptanalysis method for cryptographic chips.It mainly uses the cryptographic algorithm to leak physical information when it is executed on the chip,thereby extracting sensitive information from the cryptographic device.The power analysis method in SCA has developed rapidly and has become a powerful side-channel attack method,posing a huge threat to the security of cryptographic devices.Therefore,the ability to protect cryptographic devices from SCA has become a top priority in the information security industry.This thesis studies the side channel analysis method and implementation scheme of block cipher AES algorithm.The main work and research results are as follows:This research first introduces the principle of side channel attack,the traditional power analysis method,and the defense method of side-channel power analysis.Compared to traditional correlation power analysis,the template attack method is a very powerful attack method from the perspective of information theory.This thesis focuses on the principle and attack steps of template attack,and on this basis,the pattern recognition method(machine learning,deep learning)is applied to the field of side channel analysis.In this thesis,a power consumption acquisition and analysis platform based on AES algorithm is built,and the pre-processing method of power consumption signal is studied.Based on the research of signal filtering methods,combined with the characteristics of power consumption curves leaked by cryptographic equipment,a noise reduction method based on EEMD-wavelet transform is proposed.The related energy analysis attack(CPA)experiment is carried out by the measured power consumption signal,and compared with the traditional digital filtering method and wavelet transform method.The experimental results show that the power signal preprocessing method proposed in this thesis has better noise reduction effect and signal.The useful features are reserved more.When the signalto-noise ratio of the power consumption signal in the power analysis attack is low,the practical application is more significant.Based on the research of traditional template attack,this resarch combines principal component analysis and support vector machine(SVM)in machine learning to propose a PCA-SVM attack method and apply convolutional neural network in deep learning.In the energy analysis,a convolutional neural network(CNN)model is built.The built-in block cipher algorithm experimental platform collects the power consumption curve of AES-128 algorithm software,selects the output of the first round S box of the algorithm as the intermediate value,adopts the Hamming Weight power consumption model,and then attacks the traditional template attack and SVM class template.The model,PCASVM model and CNN model are trained and tested separately.The experimental results show that the CNN model has a much higher accuracy than the traditional template attack in the case of limited power consumption curve,which is slightly lower than the SVM class template attack,but the model algorithm has the lowest time complexity and the dependence on the number of power consumption curves.Not much.In this thesis,the AES algorithm mask defense scheme is studied,and the CNN model is applied to the energy analysis attack of the AES algorithm defense scheme.The power curve of DPA Contest V4 is experimentally verified.The power curve is the RSM mask implementation of AES-256 algorithm.Firstly,the CNN model is used to perform the mask offset offset cracking.Then,the S-box input intermediate value is used to perform the class template attack using the Hamming weight model.