Research On Hardware Trojan Detection Method Based On Side-channel Signal Analysis

With the technology of integrated circuits being more and more complex,the production cost of chips continues rising.The integrated circuit manufacturers have no competence to complete the entire IC production line on their own anymore,so that each step of IC production has to be assigned to different manufacturers to be finished.During the transfer of the IC design,the security of the IC loses its indemnity cover.This behavior leads to frequent hardware security problems which represented by hardware Trojans,and the security and reliability of integrated circuit systems are seriously threatened.During the current stage,in the face of the intricate production environment of the IC design,we have to have further study on hardware Trojans,and effectively improve hardware Trojan detection technology to provide a reliable guarantee for the security of the integrated circuit system.The hardware Trojan detection method based on side-channel analysis compares the side-channel information of the testing circuit with the golden model.According to whether there is a difference between these two part,it can be judged that the testing circuit has hardware Trojan or not.Compared with other detection technologies,this type of method has a wider application range,and the applicable circuit types are more varied,and will not cause any damage to the testing chips.However,the traditional side-channel analysis method has two drawbacks:firstly,the golden moden is needed as a reference for detection,but it is rather hard to obtain;secondly,the process variations will mask the activity of some hardware Trojans.Aiming at these two drawbacks,this paper proposes a hardware Trojan detection method based on self-similarity of circuit modules.Through comparing the same detection parameter of two structurally similar circuit modules in SoC,the difference between the detection parameters of these two circuit modules is used to realize the golden modle free detection.Meanwhile,the method selects the ratio of the dynamic current and the maximum operating frequency(IDDr/Fmax)as the detection parameter to eliminates the influence of part of the process variations on hardware Trojans activity.The software simulation based on Hspice and the hardware verification based on FPGA show that it can achieve over 90%detection success rate for both combined Trojans and sequential Trojan.For large-scale circuits,the impact of hardware Trojans on the overall circuit power consumption can not be obvious enough,so that the hardware Trojan detection method based on power analysis could not achieve the expected detection results.Therefore,considering splitting the whole circuit into each path,this paper proposes a method based on leakage power analysis to detect hardware detection by each path for reducing the influence of the process variations on the side-channel signal.Moreover,When it is determined that the testing circuit contains a hardware Trojan,the specific location of the hardware Trojan can be further located according to the detection result,such improving the detection resolution of the hardware Trojan.