Research Of Hardware Trojan Detection Technology Based On Principal Component Analysis

Hardware Trojans is the malicious circuits implanted during the IC design or manufacturing process which will disclose key information or disable the chip after being activated.Hardware Trojan detection methods are widely studied among which the side channel technique gains a lot attention.Side channel detection method acquires the differences between chips with and without Trojans and thus data processing method is a key factor for effective detection.A set of data processing recognition algorithm is proposed based on side channel analysis and PCA(principal component analysis),which is verified by power signal analysis.Firstly,power consumption information model of the implanted hardware Trojan circuit is established to get the key to achieve effective detection.Power consumption is acquired from the hardware Trojan detection platform based on FPGA after designing the circuit and optimizing testing environment and testing techniques.Secondly,the power consumption data is analyzed by using the corresponding data processing method.Data preprocessing method is used to solve the unaligned waveform,abnormal data and noise problem.What's more,the data feature extraction and feature selection is realized based on PCA and obtains over 99% of the original data information and effective detection can be achieved by distance judgment.Finally,the detection accuracy is improved after optimizing the pretreatment algorithm for hardware Trojan detection method based on PCA.And then the applicability of this method is verified by changing the parameters such as sample size,sample frequency,and hardware Trojan data processing system is developed based on MATLAB.The results show that the method based on PCA combined with distance judgment can effectively detect the Trojan circuits which account for about 0.15% of the total circuit area.