The Hardware Trojan Detection Based On Power-Channel

As the footstone of modern information devices even the information industry, IC safety has been a crucial factor to national security and profile. However, with the improvement of IC design and semiconductor manufacture process, it is easy to attack the chip and steal confidential information. In addition, for the economy profile, a large number of third party IP cores, EDA tools and foundries are used in IC design and production. Those procedures are out of the designers' control and may be inserted Hardware Trojans by attackers, which brings a big challenge to IC safety.This paper firstly introduces the concept and gives out the structure and class of Hardware Trojans. Then it sums up Hardware Trojans implantation approaches and detection methods. Afterwards, it rises up the Hardware Trojans defending ways on different levels, and analysis the security of AES algorithm circuit.We insert various size of Hardware Trojans in S-box of AES circuit and study the influence of test vectors, sample frequency, process variation, voltage jitter and temperature to detection process. Research indicates that effective test vectors can increase detection sensesitivity by 12.71% with activing trojan. When sample frequence is rose to 200 times, it can be used to compress the sample amount and increase the number of power traces. In addition, the larger fluctuation process, more than 50? temperature change and 2.5% voltage jitter can cover trojan power. Thus we put up the comprehensive Hardware Trojans detection project, which can increase the detection sensitivity and decrease the detection cost.In this paper, we mainly think about power noise such as process variation which may affect the Hardware Trojans detection. Through the power pretreatment we optimize the current mean difference detection algorithm. We also rise the power noise optimizing arithmetic based on static pretreatment and figure extract. After that we build the power matrix data analysis model with data convergence and divergence as well as projection space overlapping ratio. Experiment results show that it can effectively detect Hardware Trojans with a size of 1%.The paper takes AES algorithm circuit for example. In the experiment power analysis environment is set and we insert a function distort Hardware Trojans with a size of 1.67% in AES circuit and verify its function. Characteristic projection data which is handled by noise optimized algorithm is used in the convergence analysis and spatial overlap ratio test. The experiment results show that the hardware Trojan detection can be achieved. At last, we build ideas of validation platform of the hardware Trojan detection method.