| To provide a reliable computing environment,cryptographic algorithms are implemented in particular cryptographic devices nowadays.However,since the development of Side Channel Analysis(SCA),the implementation security of such devices has received intensive attention in the area of cryptanalysis.Recently,as a kind of postive SCA,fault injection analysis brings a severe threat to cryptosystems because it allows attackers to interfere with the process of cryptographic algorithms postively.In 2018,Zhang et al.proposed Persistent Fault Analysis(PFA).Persistent fault is a novel fault which falls between transient fault and permanent fault.The attackers are able to recover the key with only the ciphertexts of the cryptographic device by injecting persistent faults into the cryptographic device.PFA has shown so great potential that it is robust against the classical fault injection analysis countermeasures,such as Dual Modular Redundancy(DMR)or mask.On the other hand,there are still some shortages in the studies of PFA.The attackers have to know the memory values which are altered by persistent faults in the first.Secondly,to improve the success rate of PFA,a large mount of ciphertexts are required.Besides,it has become an urgent problem in the area of fault injection analysis to find a countermearsure which is able to prevent PFA efficently.To solve these problems,the main works of this thesis are as follows:Firstly,to push the limit of PFA,we propose a performance optimization anaginst PFA.We combine Statistical Persistent Fault Analysis(SPFA)and Enhanced Persistent Fault Analysis(EPFA)to Enhanced Statistical Persistent Fault Analysis(ESPFA),which is more efficient.To bypass the requirement of the memory values,we exploit the distribution character of cryptographic algorithm which is influenced by persistent faults and compare the value of the Squared Euclidean Imbalance(SEI)in each key guess.Besides,we reduce the number of ciphertexts which are used to recover the key by exploiting the fault leakage in more rounds.Then,to provide an efficient countermearsure against PFA,we propose a fastdetection and fault-correction algorithm.We construct a fixed input and output pair to detect faults rapidly.Then we build redundant tables to store the relationship between the adjacent elements in the S-box,by which the algorithm can correct the faulty elements in the S-box.Our algorithm can prevent PFA effctively by keeping the correctness of ciphertexts.We conduct experiments with AES in both software implementation in μC and hardware implementation in FPGA for our performance optimization and countermearsure.The experimental results demonstrate the feasibility and efficiency of our performance optimization and countermearsure. |
