论文部分内容阅读
硬件木马成为近年来硬件安全领域的一个新威胁。硬件木马可能在集成电路设计的各个阶段插入到电路中,一经流片,就不可修复更改,将对通信和安全等领域造成巨大损失。针对时序型硬件木马,提出一种不需要基准芯片的自比较硬件木马检测技术。即通过对电路添加相同重复的激励,比较对应激励在不同时间窗口的电流值,判断电路中是否含有木马。将高级加密标准(AES)加密电路在SMIC 0.18μm,0.13μm,65 nm,55 nm四种工艺库条件下进行仿真,并结合马氏距离处理实验数据,增加检测的区分度。仿真数据区分度提高了240.6倍,现场可编程门阵列(FPGA)实测数据增加了3.5倍。实验结果表明,基于自比较的硬件木马检测技术能够检测出比AES电路小2个数量级的硬件木马。
Hardware Trojans have become a new threat in the field of hardware security in recent years. Hardware Trojans may be plugged into the circuit at all stages of the IC design. Once the chip is removed, the changes can not be repaired, which can cause huge losses in areas such as communications and security. Aiming at the time-series hardware Trojan, a self-comparing hardware Trojan detection technology which does not need a reference chip is proposed. That is, by adding the same repetitive stimulus to the circuit, comparing the current value of the corresponding stimulus at different time windows to determine whether the circuit contains a trojan. The AES encryption circuit was simulated under SMIC 0.18μm, 0.13μm, 65 nm and 55 nm technology library, and the experimental data were combined with Mahalanobis distance to increase the discrimination of the detection. The discrimination of simulation data is increased by 240.6 times, and the measured data of field programmable gate array (FPGA) is increased by 3.5 times. The experimental results show that hardware Trojan detection based on self-comparison can detect hardware Trojans that is 2 orders of magnitude smaller than the AES circuit.