A Self-Learning Structure to Improve the Efficiency of BlueChip Software

There have been serious concerns about hardware security from malicious modification of hardware design during design or manufacturing. Several hardware Trojan detection approaches have been proposed in past few years. In [1], Hicks et al. put forward a novel strategy named BlueChip to detect and nullify the hardware Trojans at design time. They also proposed a recovery mechanism in software layer to assist the modified hardware work. This detection and resistance method is useful while it has some flaws. Some legitimate circuits which were removed by BlueChip might invoke the recovery function continuously. In this paper, we propose a novel approach to improve the effectiveness of the BlueChip architecture by building a self-learning structure to let the circuit distinguish the removed legitimated circuits from the removed Trojan circuits.