Hardware Trojan Detection Optimization Based On Wavelet De-noising Data Preprocessing

Hardware Trojan are malicious redundant circuits which can be implanted in the circuits from any link of IC designing and manufacturing,once activated under certain condition,it could change the circuits original function or disclose key information or disable the chip.Because of the emergence of hardware Trojan circuit has seriously threatened the information security,therefore hardware Trojan methods are widely studied,and the side channel analysis technology is the most widely studied hardware Trojan detection technology.This paper mainly focuses on the research of the side channel analysis technology,which is verified by power signal analysis.The paper aiming at the problem of poor data preprocessing in hardware Trojan detection,the optimized hardware Trojan detection method of wavelet transform of the noise reduction data preprocessing is proposed.In this paper the existing hardware Trojan detection technology are introduced,and expounds the types and principle of side channel analysis technology.According to the characteristics of power consumption information,the carrier circuits and hardware Trojan circuits are designed.Afterwards,ISCAS'89 series benchmark circuits containing Trojans are implemented on FPGA,then the power information is processed,and obtain the power consumption information of the reference circuit and the embedded hardware Trojan horse.In this paper the wavelet transform of the noise reduction process is introduced.Based on the experiments of different wavelet functions,decomposition levels and threshold processing are carried out on the same set of power consumption data,the article use wavelet function sym8 to reduction the noise of power consumption data,At the same time,the power data is decomposed by three layers of wavelet decomposition,then using Rigorous SURE threshold value to deal with the high frequency coefficient which obtained by wavelet decomposition.The paper process the power consumption information by Markov distance to identify hardware Trojan,which is preprocessed by wavelet de-noising.Through the contrast experiments that the same set of power consumption data is processed by Markov distance and dimension processed by PCA,the article obtain the result that the data of wavelet de-noising preprocessing method is high-efficiency.The experimental results show that the wavelet de-noising preprocessing method can detect hardware Trojan with an area ratio of 0.24 %.