Research And Implementation Of Side Channel Attack Method For Cryptographic Chips

Mathematically secure cryptographic algorithms are challenged by side channel attacks during the operation of physical devices.There are two main ways to implement cryptographic algorithms on these devices.One is the hardware implementation with coprocessor as auxiliary unit.The common representatives are ASIC and FPGA implementation.The other is the software implementation based on CPU directly.The common representatives are various microcontrollers.Due to different side channel information leakage of different implementations,the research work of this thesis will be based on two implementations of cryptographic chips.For software implementation,this thesis focuses on template attack.Firstly,three steps of template attack are described,including feature extraction,template construction and template matching.Secondly,aiming at the shortcomings of traditional feature extraction methods,this thesis proposes a feature extraction method based on decision tree.Experiments show that this method can achieve a high attack accuracy under a small number of feature points.Under unsupervised conditions,this thesis introduces the Gauss mixture clustering algorithm,and improves the classification and attack strategy,so that template attacks can still be implemented.Subsequently,during the attack of a smart card implemented by mask software,machine learning algorithm is compared with template attack method.Experiments show that ensemble learning is superior to others.For hardware implementation,this thesis focuses on the differential power analysis.For differential power analysis attacks,this paper uses data standardization to reduce false peaks.For mutual information analysis attacks,this thesis uses the assumptions of relative entropy and normal distribution,so that only variance information is needed to implement attacks.For the correlation power analysis attack,this thesis compares its attack effect under the principal component analysis and frequency domain analysis.The experiment shows the attack under the frequency domain analysis is better.In addition,a differential power analysis attack with multi-register leakage is proposed from the perspective of registers.Experiments show that this method can greatly reduce the number of energy traces for a successful attack.In order to facilitate research,this thesis designs a side channel attack verification platform,which mainly includes power acquisition and import module,data preprocessing module and attack module.The platform can realize the functions of communication with encryption chip,real-time acquisition of power consumption data and verification of attack algorithm.It makes the process of side channel attack more concise and lays a foundation for further research of attack algorithm in the future.