| As the foundation of the information society,integrated circuits have been applied to many industries.The development of the integrated circuit industry has been listed as a strategic goal of China.However,the information security incidents that have been exposed in recent years have brought tremendous impact to the security of the integrated circuit industry.As a major threat,hardware Trojans are gradually paid much attention.And the technology of hardware Trojan detection has become a hot research area.In order to comprehensively analyze feature of hardware Trojan and achieve precision detection,this paper starts from the perspective of the attacker to study the implantation,attack and activation modes of the hardware Trojan.AES,DES,RS232 and I2 C circuits are selected as the golden reference circuits,fourteen kinds of RTLlevel hardware Trojan were designed and simulated,and functions such as information leakage,function change,performance reduction and denial of service are realized.In addition,the layout of the electromagnetic leakage hardware Trojan AES-T1 was designed and taped out successfully,the demonstration platform was built and the hardware Trojan attack demonstration was completed at the chip level.Then,the basic characteristics such as physical properties and concealment of the hardware Trojan are analyzed and summarized.On this basis,by studying the relationship between hardware Trojan concealment and circuit activity,the Trojan Activation Module(TAM)is designed from the perspective of circuit structure.When activated,the module can improve the overall activity of the circuit to the maximum extent by changing the logic operation mode of the circuit,so as to enhance the activation degree of hardware Trojan and reduce its concealment.The validity of the structure was verified by implantation and detection experiments.Finally,based on the feature of side channel,starting from the two aspects of power consumption side channel feature and electromagnetic side channel feature,the composition and influence factors of the two types of information are deeply analyzed.Four hardware Trojans are designed based on AES in the previous article are selected as the experimental objects,and the two types of feature information are analyzed,collected and preprocessed.The RBF neural network was selected,constructed and trained as a recognition algorithm to identify the two types of feature information.For the four hardware Trojans,the detection rates of power consumption information are 98.9%,97.7%,97.7%,and 98.1%,respectively,the detection rate of electromagnetic information is better than the power consumption information,which are 100%,99.0%,99.2%,and 98.5%,respectively.By extracting and identifying side channel feature information,this paper successfully implemented high-precision hardware Trojan detection.And the effectiveness of hardware Trojan side channel feature recognition technology based on RBF neural network are verified. |