基于IFCM加权的SVDD硬件木马检测方法

Hardware Trojan detection method based on IFCM weighted SVDD

针对硬件木马(HT)种类繁多难以获取未知木马特征及采集的旁路信号含噪声问题,提出了一种基于IFCM加权的SVDD(IFCMW-SVDD)硬件木马检测方法。传统支持向量数据描述(SVDD)在解决单分类问题时存在相同条件下训练全部样本的不足,需要根据相应问题对样本有主次之分进行训练。通过一种改进的模糊C均值方法(IFCM)计算金片旁路信号的隶属度,将其作为样本特征的权重(W)系数,使得针对硬件木马检测问题构建SVDD模型的支持向量能够描述金片信号的同时尽可能减小描述范围。实验表明,所提方法实现单分类硬件木马检测的同时较传统SVDD算法在检测精度和稳定性上都有所提高。

Aiming at the problems that a great variety of hardware Trojans,it is difficult to obtain unknown Trojan features,and the collected side-channel signals contain noise,this paper proposed a hardware Trojan(HT)detection method based on IFCM weighted SVDD(IFCMW-SVDD).The traditional SVDD has the defective of training all the samples under the same conditions when solving the single classification problem,samples need to be trained in order of priority according to the corresponding problems.But this algorithm calculated the membership degree of the golden chip bypass signal by IFCM,and used it as the weight(W)coefficient of the sample feature,the support vectors of constructed SVDD model for the hardware Trojan detection problem could describe the golden chip signals while minimizing the description range.Experiments show that the proposed method achieves the detection of single-class hardware Trojans and has higher detection accuracy and stability than the traditional SVDD algorithm.