The influence on sensitivity of hardware Trojans detection by test vector

Hardware Trojans have emerged as a new threat to the security of chip and the reliability of computing systems. As such, methods of determining whether a “security chip” contains hardware Trojans are a research concern. This study proposes power relative variation (PRV) parameters to analysis the relation between test vectors and detection sensitivity. Following a brief introduction into Trojans and detection methods, we present the S-box circuit noise model in AES core and simulate the power tracks of various sizes of hardware Trojans using HSPICE. We also apply different sets of test vectors in 40 nm processing. Experimental results show that the power of the entire circuit is reduced by 41% and the PRV value increased 12.71% and 3.34% at most corresponding to combinational and sequential Trojan circuits. The valid test vectors can enhance the activity of hardware Trojans. They can also limit the noise from process, supply voltage, and temperature and extrude the features from the Trojan circuit to increase PRV value. As a result, a sample amount can be compressed at a reduced cost given a similar detection effect. Hence, the investigation into the influence of test vectors on the detection methods of hardware Trojans based on side-channel analysis is extremely important.