基于朴素贝叶斯分类器的硬件木马检测方法

发布时间：2018-05-24 08:15

  本文选题：侧信道分析 + 硬件木马　； 参考：《计算机应用研究》2017年10期

【摘要】：在侧信道分析的基础上,针对芯片中存在的硬件木马,提出一种基于朴素贝叶斯分类器的硬件木马检测。该方法能够利用训练样本集构建分类器,分类器形成后便可将采集到的待测芯片功耗信息准确分类,从而实现硬件木马检测。实验结果表明,对于占电路资源1.49%和2.39%的两种木马,贝叶斯分类器的误判率仅为2.17%,验证了该方法的有效性和适用性。此外,在与欧氏距离判别法比较时,基于朴素贝叶斯分类器的方法表现出了更高的判别准确率,同时也具有从混杂芯片中识别出木马芯片与标准芯片的能力,这又是马氏距离判别法所不具备的。
[Abstract]:On the basis of side channel analysis, a hardware Trojan horse detection based on naive Bayes classifier is proposed for the hardware Trojan horse in the chip. This method can use the training sample set to construct the classifier. After the classifier is formed, the power consumption information of the chip to be tested can be accurately classified, thus the hardware Trojan can be detected. The experimental results show that the Bayesian classifier's error rate is only 2.17 for the Trojan horse which accounts for 1.49% and 2.39% of circuit resources. The validity and applicability of this method are verified. In addition, compared with Euclidean distance discriminant, the method based on naive Bayesian classifier shows higher accuracy, and it also has the ability to recognize Trojan and standard chips from hybrid chips. This is again the Markovian distance discriminant method does not have.
【作者单位】： 北京电子科技学院电子信息工程系;
【基金】：中央高校基本科研业务费专项资金资助项目(2014GCYY04)
【分类号】：TN407
，

本文编号：1928342