基于温度特征分析的硬件木马检测方法

Hardware Trojan Detection Through Temperature Characteristics Analysis

硬件木马是一种在特定条件下使集成电路失效或泄露机密信息等的恶意电路,给现代信息系统带来了严重的安全隐患。该文基于硬件木马在芯片工作之初造成的温度响应特征,提出一种利用芯片温度变化特性并进行比对的硬件木马检测方法。该方法采用环形振荡器作为片内温度特征测量传感器,提取温度变化特征信息,并采用曲线拟合评价指标来评估硬件木马对温度变化特征的影响,通过比对无木马芯片温度响应特征从而完成木马检测。通过对10个不同芯片的检测,结果表明该方法能够对面积消耗32个逻辑单元硬件木马的检测率达到100%,对16个逻辑单元检测概率也能达到90%;同时检测结果表明该方法完成硬件木马检测后,能够对硬件木马的植入位置进行粗定位。

Hardware Trojan is the malicious circuit modification which can disable the Integrated Circuit（IC） or leak confidential information covertly to the adversary, and brings potential safety hazard for ICs. In this paper, a new approach for hardware Trojan detection based on compare the temperature variation characteristics when IC starts working. Ring Oscillator（RO） is used as a detector to obtain the information about IC＇s temperature variation characteristics. In order to describe temperature variation characteristics accuracy, a parameter about the D-value of RO＇s oscillation cycle counts is presented, and parameters about the quality of the fitting curve are used to estimate the hardware Trojan＇s effect on IC＇s temperature characteristics. Results from ten chips show that the proposed approach is effective towards increasing successful detection ratio and can achieve better Trojan detection probability 100% on average over conventional patterns for Trojan which is 32 logic elements, and for Trojan which is 16 logic elements can also achieve Trojan detection probability 90%, besides the proposed approach locating the Trojan＇s insertion place roughly.