Securing FPGAs in IoT: a new run-time monitoring technique against hardware Trojan

In this paper, we propose a new run-time monitoring technique to detect hardware Trojan (HT) on working field-programmable gate array (FPGA), which is helpful for reinforcing the hardware security of IoT devices. First, we place multiple temperature sensors on FPGA and collect their value at each time interval. Then, we construct a predictive model which can estimate the distribution of sensor frequency in real-time. If the difference between the predicted value and the real one surpasses a dynamically updated threshold, an HT activation is reported. The experiment results reveal that our technique has high detection accuracy (≥ 98%), low false negative ratio (≤ 1.5%) and zero false positive ratio under various FPGA HT benchmarks.