Hardware Trojan Detection using Power Signal Foot Prints in Frequency Domain

This work proposes a plausible detection scheme for Hardware Trojan (HT) detection in frequency domain analysis. Due to shrinking technology every node consumes low power values (in the range of μW) which are difficult to manipulate for HT detection using conventional methods. The proposed method utilizes the time domain power signals which is converted to frequency domain that represents the implausible signals and analyzed. The precision of HT detection is found to be increased because of the magnified power values in frequency domain. This work uses ISCAS89 bench mark circuits for conducting experiments. In this, the wide range of power values that spans from 695 μW to 22.3 μW are observed in frequency domain whereas the respective powers in time domain have narrow span of 2.29 μW to 0.783 μW which is unconvincing. This work uses the wide span of power values to identify HT and observed that the mid-band of frequencies have larger footprints than the side bands. These methods intend to help the designers in easy identification of HT even of single gate events.