Research On Fault Model And Countermeasure Effectiveness Evaluation Of Clock Glitch Fault Injection Attack

Information security problems have become acute in recent years.Although cryptographic algorithms provide some protection for critical information,integrated circuits implementing cryptographic algorithms are vulnerable.Attacks on cryptographic circuits mainly include side channel attack and fault injection attack.The latter firstly actively injects faults into cryptographic circuits,and then processes the faulty results by fault analysis methods to retrieve the key.Fault injection attack has strong attack capability and high attack efficiency.However,the fault models proposed by most fault analysis methods are strict.Using the existing fault injection techniques to achieve these strict fault models will result in high cost or a large amount of useless fault injections.For example,clock glitch fault injection technique is low-cost but lowaccurate,and available faults require to be searched from a large number of fault injections.Therefore,in order to reduce the cost of fault injection attack and improve the efficiency of fault injection attack,low-cost clock glitch fault injection technique and its associated fault analysis methods deserve to be carefully studied.In this paper,the fault behavior on hardware circuits caused by clock glitch fault injection technique is analyzed in detail,and single fault model and multi-fault model related to clock glitch fault injection are built.Based on these two fault models,the existing fault analysis method with multiple plaintexts and single fault(MPSF)and another existing fault analysis method with single plaintext and multiple faults(SPMF)are improved.The enhanced fault analysis methods are proposed,namely eMPSF analysis and eSPMF analysis.Four fault analysis methods are applied to AES(Advanced Encryption Standard)cryptographic circuits through theoretical attack and simulated attack.The experimental results show that the improved fault analysis methods are superior in terms of attack efficiency and attack success rate.In particular,eSPMF analysis can increase the attack success rate of SPMF analysis from 84.375%to 96.25%,but the average number of faults required to recover each key byte is reduced from 2.7037 to 2.4740.On the other hand,the development of fault injection attacks promotes various countermeasures to be proposed,thus,the evaluation methods for measuring the effectiveness of countermeasures need to be studied.This paper proposes an information theory-based evaluation method,which uses the security factor comparing the theoretical amount of information leakage and the actual amount of information leakage as metric to quantify the effectiveness of countermeasures in cryptographic circuits with SPN(Substitution Permutation Network)structure,and is applicable to various fault models.In this paper,the theoretical amount of information leakage under the single fault model and multi-fault model related to clock glitch fault injection is derived and three methods of calculating the actual amount of information leakage are proposed.This quantitative evaluation method is applied to AES cryptographic circuit.The experimental results demonstrate that the effectiveness of countermeasures can be judged by the level of security factor.This evaluation method is feasible.In addition,the comparison of three calculation methods for actual amount of information leakage indicates the latter two effective calculation methods can optimize the number of fault injections and the time required for evaluation.Finally,under two single fault models and one multi-fault model,the evaluation time required for the effective calculation methods is always less than 0.01 s,which indicates the effectiveness of this evaluation method.