基于概率分布的硬件木马检测特征选择方法

针对基于旁路信号分析的硬件木马检测中存在的高维以及信号冗余度高等问题,寻求一种特征选择方法在降维、降低冗余度方面的可行性,通过考虑样本概率的分布情况,提出了一种以散度和Bhattacharyya距离作为可分性判据的特征选择方法。首先分析了旁路信号的特征选择问题,然后阐述了基于散度和Bhattacharyya距离的可分性判据,最后在FPGA中植入硬件木马,采用K-L方法进行实验,通过对样本特征选择前后的检测效果发现,这种特征选择方法不仅有助于分辨"金片"与含木马的待测芯片间旁路信号的统计差异,而且与分类错误率建立联系,更好地实现了硬件木马的检测。

Hardware Trojan Detection Feature Selection Method Based on Probability Distribution

Aiming at the existing problems such assignal redundancy and high dimension problems in Hardware Trojan detection based on the side-channel analysis,to explore a feasibility of feature selection method in removing redundancy and reducing of side-channel signal,by considering the probability distribution of samples ,then a feature selection method using Divergence and distance of Bhattacharyya as class divisible criterion of feature selection method is put forward.First,the feature selection problem in side-channel,then elaborated the calss separability criterion based on Divergence and distance of Bhattacharyya are analyzed,In the end,we implant the Trojans into the FPGA chip,and test it using K-L method,through the comparison before and after the detection effection side-channel signal feature selection we can see,this method can help to distinguish side-channel signal statistical characteristic difference between"gold chip"and Trojan chip(contain Trojan),establish direct relations with classification error rate,and better implements the Hardware Trojan Detection.