Hardware Trojan Detection Technology Based On Power Side Channel

In the wake of developments in electronic information technology,integrated circuit chips are widely used in military and people's livelihoods,and play an important role in national and social security.The industrial supply chain model in which integrated circuit design and production are separated,and each module design is independent of each other,makes the chip inevitably have hardware security risks.A hardware Trojan horse refers to a structure that intentionally modifies or implants a circuit during the design or manufacturing stage of an integrated circuit,causing malicious behavior in the circuit.Hardware Trojan horses are very harmful,so the research on hardware Trojan horse detection technology can ensure chip security,which has important significance for national and social security.First,This article builds a hardware Trojan horse circuit library based on the AES encryption circuit,and the Trojan horse circuit in the library is used as the detection target.The Trojan horse detection technology based on the power consumption side channel needs to compare the power consumption waveforms of the reference circuit and the circuit to be tested.In order to collect the power consumption waveforms of the circuit,an FPGA-based power consumption acquisition platform is built.The platform is equipped with an amplifier.Acquire the normal or amplified power consumption waveform of the circuit.Finally,this article uses two hardware Trojan detection methods.First,the principle component analysis method combined with the Mahalanobis distance method is used to detect the Trojan horse circuit.When using the principle component analysis method to reduce the original data dimension,part of the feature information will be lost,which leads to this method.The detection accuracy of small hardware Trojans is poor.In order to further improve the detection effect of small Trojan horses,this paper proposes to use BP neural network to directly extract the characteristic information of circuit power consumption waveforms to detect hardware Trojan horses,and by amplifying the power consumption waveforms of small Trojan horse circuits to determine the power consumption difference between the two types of circuits,Further improve the detection accuracy of BP neural network.It can be seen from the experimental results that the BP neural network detection method can effectively improve the detection accuracy of hardware Trojan horses.The detection method of amplified power waveform combined with BP neural network can achieve 100%detection accuracy for Trojan horse circuits with an area ratio of greater than 0.1%,and 99.301%for Trojan horse circuits with an area ratio of 0.089%.