硬件木马的边信道提升技术研究

发布时间：2018-04-15 10:34

本文选题：硬件木马检测 + 边信道分析　；参考：《电子科技大学》2016年硕士论文

【摘要】：硬件木马问题已成为信息安全防护中必须考虑的重要问题,随着集成电路的迅速发展,芯片已经应用到人类生活的各个方面,因此,保证芯片的安全对国家的政治经济发展、国防建设和百姓的信息财产安全都至关重要。因此,开展基于边信道分析的硬件木马检测技术意义重大。本文重点研究了不同工艺偏差下基于边信道分析的硬件木马检测方法,并提出了可以降低工艺噪声和数据冗余的功耗数据处理方法,这种方法可以有效提高木马电路的信噪比,同时对提高木马激活概率的方法展开了深入研究,用以辅助边信道分析技术检测。本文研究的主要内容如下:1.模拟不同工艺偏差对电路的影响。研究了工艺偏差对电路功耗和延时的影响,通过修改工艺库的高斯参数,设置了三种工艺偏差,对应的电流偏差(Current Variation,CV)分别为±2%,±5%和±7%,其中±5%是SMIC 130nm工艺库提供的最大偏差。采用基于蒙特卡罗仿真的工艺偏差模拟方法对四种电路和对应的木马电路进行平均瞬态电流分析,当CV为±7%时,可以检测的最小木马面积为2%;CV为±5%时,可以检测的最小木马面积为1.5%;CV为±2%时,可以检测的最小木马面积为0.8%,检测效率很低,说明了工艺偏差的存在对硬件木马的检测产生了极大的干扰。2.基于功耗分析的硬件木马检测的数据处理方法研究。建立了适用于信号特征提取的木马检测模型,重点研究了K-L变换和马氏距离分析的数据降维降噪方法,两种方法进行结合可以有效提高木马电路的信噪比。实验结果表明,与未处理的功耗数据相比,处理后的原始电路数据和木马电路数据分离度明显提高,当CV为±7%时和±5%时,可以检测的最小木马面积为0.6%;CV为±2%时,可以检测的最小木马面积为0.34%。3.提高硬件木马激活概率的方法研究。通过分析影响木马激活概率的因素,在电路中插入一种可以提高木马激活概率的结构,并改进了该结构的插入算法,减少了该结构对电路面积的影响,最大面积优化率达到44.8%。经仿真和功耗分析证明,在电路面积增加不超过0.44%的前提下,该结构可使电路所有节点的翻转率提高到阈值(thP)以上,增大了木马电路内部节点的翻转概率,增大了木马电路被激活的概率,缩短了木马电路的激活时间,使得木马电路功耗对电路总功耗的贡献率平均提高了0.39%,从而有效辅助基于边信道分析的硬件木马检测技术研究。
[Abstract]:Hardware Trojan has become an important problem in information security protection. With the rapid development of integrated circuits, chips have been applied to all aspects of human life. Therefore, to ensure the security of chips is of great importance to the political and economic development of our country.National defense construction and people's information property security are very important.Therefore, it is of great significance to develop hardware Trojan detection technology based on side channel analysis.This paper focuses on the hardware Trojan detection method based on side channel analysis under different process deviations, and puts forward a power consumption data processing method which can reduce process noise and data redundancy. This method can effectively improve the signal-to-noise ratio of Trojan circuit.At the same time, the methods to improve the activation probability of Trojan horse are studied in depth to assist edge channel analysis technology detection.The main contents of this paper are as follows: 1.The influence of different process deviation on the circuit is simulated.The effect of process deviation on circuit power consumption and delay is studied. By modifying Gao Si parameters of process library, three kinds of process deviations are set up. The corresponding current deviation is 卤2, 卤5% and 卤7 respectively, and 卤5% is the maximum deviation provided by SMIC 130nm process library.The method of process deviation simulation based on Monte Carlo simulation is used to analyze the average transient current of four circuits and corresponding Trojan circuits. When CV is 卤7, the minimum Trojan horse area can be detected when the CV is 卤5.When the minimum Trojan horse area can be detected is 1.5 and CV is 卤2, the minimum Trojan horse area can be detected is 0.8 and the detection efficiency is very low, which shows that the existence of process deviation has great interference on the detection of hardware Trojan horse.Research on data processing method of hardware Trojan Horse Detection based on Power Analysis.A Trojan horse detection model suitable for signal feature extraction is established. The data dimension reduction and noise reduction method based on K-L transform and Markov distance analysis are studied. The combination of the two methods can effectively improve the signal-to-noise ratio of Trojan horse circuit.The experimental results show that compared with unprocessed power consumption data, the separation degree of original circuit data and Trojan circuit data after processing is obviously improved. When CV is 卤7 and 卤5, the minimum Trojan horse area can be detected when the CV is 卤2.The minimum area of the Trojan that can be detected is 0. 34 and 3. 3.Research on the method of improving the activation probability of hardware Trojan horse.By analyzing the factors that affect the activation probability of Trojan horse, a structure that can increase the activation probability of Trojan horse is inserted into the circuit, and the insertion algorithm of the structure is improved, which reduces the influence of the structure on the circuit area, and the maximum area optimization rate reaches 44.8%.The simulation and power analysis show that under the premise of increasing the circuit area by less than 0.44%, the structure can increase the turnover rate of all nodes of the circuit to the threshold value and increase the turnover probability of the inner nodes of the Trojan circuit.It increases the probability of Trojan circuit to be activated, shortens the activation time of Trojan horse circuit, and increases the contribution rate of Trojan circuit power consumption to the total power consumption of circuit by an average of 0.39, thus effectively assisting the research of hardware Trojan horse detection technology based on edge channel analysis.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP309

【参考文献】