Research And Implementation Of Power Analysis Based On Symmetrical Cryptography

As the core of information security,cryptography has been concerned by more and more researchers.Side channel analysis(SCA),as a new password cracking technology,finds another way.Its basic idea is to obtain some key physical information,such as voltage and electromagnetic information,during the operation of cryptographic equipment,and further analyze these physical information to obtain.The emergence of side channel attacks brings new threats to information security,and power analysis attack is the most threatening one of side channel attacks.This thesis mainly studies the power analysis attack against symmetric encryption algorithm SM4.The main work contents are as follows.This thesis first introduces the development history and research background of the subject,then describes the principle of SM4 encryption algorithm in detail,then introduces the energy consumption related knowledge of cryptographic equipment and common energy simulation models,and also introduces some traditional power consumption analysis methods.In this thesis,the low-pass filter is used to reduce the noise of the collected curve.However,the low-pass filter simply filters out part of the electronic noise.Therefore,this thesis proposes a noise reduction method based on energy trace reconstruction,and compares this method with the traditional low-pass filtering method.It is found that the c index and c index of the curve based on reconstructed energy trace are better than the traditional filtering method,which can remove redundant information more effectively.In addition,in template attack,too many feature points may lead to matrix ill condition,and too many redundant points may lead to template distortion and reduce the success rate of matching;Therefore,through the horizontal and vertical analysis of the traditional SOD algorithm,this thesis proposes an improved feature extraction method.Combined with the above energy curve preprocessing methods,the power analysis method based on SM4 algorithm is deeply studied in this thesis.Firstly,the traditional template attack is implemented for SM4 algorithm.Combined with some occasions in practice where the encrypted content of cryptographic equipment cannot be obtained at will,combined with the relatively independent characteristics of key expansion and encryption process in SM4 algorithm,this thesis carries out template attack on the power consumption curve when the key and system function are XOR in key expansion.In addition,by analyzing the internal relationship between cryptography and machine learning,class template attacks against F function and key extension function of SM4 algorithm are proposed respectively.Based on template attack,this method introduces machine learning method,takes the middle value or middle value Hamming weight in the encryption process as the classification label,and takes the trained machine learning network model as the classifier to predict the middle value or middle value Hamming weight of the curve to be attacked.In this thesis,convolutional neural network model and random forest model are used as classifiers for class template attacks.Finally,through comparison,it is found that class template attack has a higher attack success rate than traditional template attack.