基于温度传感器的硬件木马检测方法  

A Hardware Trojan Detection Method Based on Temperature Sensor

作　　者：黄正峰[1] 孙芳 潘尚杰 欧阳一鸣[2] 倪天明 戚昊琛[1] 徐奇 HUANG Zhengfeng;SUN Fang;PAN Shangjie;OUYANG Yiming;NI Tianming;QI Haochen;XU Qi(School of Electronic Science and Applied Physics,Hefei Univ.of Technology,Hefei 230009,P.R.China;School of Comput.and Inform.,Hefei Univ.of Technology,Hefei 230009,P.R.China;College of Elec.Engineer.,Anhui Polytechnic Univ.,Wuhu,Anhui 241000,P.R.China)

机构地区：[1]合肥工业大学电子科学与应用物理学院,合肥230009 [2]合肥工业大学计算机与信息学院,合肥230009 [3]安徽工程大学电气工程学院,安徽芜湖241000

出　　处：《微电子学》2020年第6期920-925,931,共7页Microelectronics

基　　金：国家自然科学基金资助项目(61874156,61874157,61904001,61904047);安徽省自然科学基金资助项目(1908085QF272)。

摘　　要：针对硬件木马检测的旁路信号分析法中需要黄金模型、受工艺扰动影响大的问题,提出了一种基于温度传感器的硬件木马检测方法。采用抗工艺扰动设计使温度传感器受工艺扰动的影响程度低。将温度传感器植入芯片内部相似结构(存储单元、功能相同的模块等),读取温度传感器的频率信息,通过简单异常值分析法与差值分析法比对相似结构的频率差异,实现了硬件木马的检测。该方法既有效克服了工艺扰动的影响,又不需要黄金模型。温度传感器输出频率在最极端工艺角下的工艺扰动仅为9%。在SMIC 180 nm CMOS工艺下对高级加密标准(AES)电路的木马检测进行了验证,结果验证了该方法的有效性。Aiming at the problem that the side-channel signal analysis method of hardware Trojan detection needed golden model and was greatly affected by process variation, a hardware Trojan detection method based on temperature sensor was proposed. The anti-process variation design was adopted to make the temperature sensor less affected by the process variation. Temperature sensors were inserted into similar structures(memory units, modules with same function, etc.) of the chip to simulate the temperature. Simple outlier analysis and difference analysis were used to compare the frequency of the temperature sensor between the similar structures to detect hardware Trojan. This method could effectively overcome the influence of process variation without requiring a golden model. The frequency of the temperature sensor affected by process variation was 9% at the worst corner. Based on SMIC 180 nm CMOS process, the hardware Trojan detection of advanced encryption standard(AES) circuit was simulated and verified.

关 键 词：硬件木马 温度传感器 相似结构 工艺扰动 

分 类 号：TN407[电子电信—微电子学与固体电子学]