Hardware Trojan Detection Method Based on Time Feature of Chip Temperature

Abstract

Hardware security has become a security issue that cannot be ignored in the world. Academia's research on hardware Trojan detection at IC (Integrated Circuit) has been ongoing and has achieved many achievements. At present, some scholars are exploring hardware Trojan detection based on chip power and temperature. In this paper, we also use the temperature information of the chip to carry out hardware Trojan detection, but the difference is that the existing methods directly used the difference of steady-state temperature after the chip reaches steady state as the detection basis, and our method is to use the arrival time difference in the temperature rising phase of the chips before reaching the steady state as the detection basis, then extract the feature to obtain the characteristic value of each chip, and finally use them to perform classification and identification of infected chips. Our basis could show larger difference for the infected chips under the same conditions, which makes the detection data have better accommodation for measurement noise and process variation (PV) and improves the detection accuracy. Through the relevant experimental verification, the proposed method is completely feasible, and still has high detection accuracy with measurement noise and PV. The proposed method can be used in the case of Trojan detection for a certain number of chips without golden chip.