CN112163245A - Hardware Trojan horse detection method based on weighting parameter Mahalanobis distance - Google Patents

Hardware Trojan horse detection method based on weighting parameter Mahalanobis distance Download PDF

Info

Publication number
CN112163245A
CN112163245A CN202011028757.3A CN202011028757A CN112163245A CN 112163245 A CN112163245 A CN 112163245A CN 202011028757 A CN202011028757 A CN 202011028757A CN 112163245 A CN112163245 A CN 112163245A
Authority
CN
China
Prior art keywords
sample
mahalanobis distance
trojan horse
hardware trojan
matrix
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202011028757.3A
Other languages
Chinese (zh)
Inventor
黄姣英
李胜玉
高成
刘基强
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beihang University
Original Assignee
Beihang University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beihang University filed Critical Beihang University
Priority to CN202011028757.3A priority Critical patent/CN112163245A/en
Publication of CN112163245A publication Critical patent/CN112163245A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
    • G06F21/70—Protecting specific internal or peripheral components, in which the protection of a component leads to protection of the entire computer
    • G06F21/71—Protecting specific internal or peripheral components, in which the protection of a component leads to protection of the entire computer to assure secure computing or processing of information
    • G06F21/76—Protecting specific internal or peripheral components, in which the protection of a component leads to protection of the entire computer to assure secure computing or processing of information in application-specific integrated circuits [ASIC] or field-programmable devices, e.g. field-programmable gate arrays [FPGA] or programmable logic devices [PLD]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F15/00—Digital computers in general; Data processing equipment in general
    • G06F15/76—Architectures of general purpose stored program computers
    • G06F15/78—Architectures of general purpose stored program computers comprising a single central processing unit
    • G06F15/7803—System on board, i.e. computer system on one or more PCB, e.g. motherboards, daughterboards or blades
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Hardware Design (AREA)
  • Theoretical Computer Science (AREA)
  • General Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Computing Systems (AREA)
  • Microelectronics & Electronic Packaging (AREA)
  • Mathematical Physics (AREA)
  • Computer Security & Cryptography (AREA)
  • Software Systems (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

A hardware Trojan horse detection method based on a weighting parameter Mahalanobis distance is divided into 5 steps. The method specifically comprises the steps that a first pair of standard and FPGA to be tested is used for collecting port power consumption under the normal operation of a direct-current power supply, and a standard and sample to be tested are obtained. The second step is to calculate and classify the traditional Mahalanobis distance of the collected power consumption data, and the third step is to select the significance level of the conventional Mahalanobis distance value of the sample and judge whether the hardware Trojan horse exists or not. And if no obvious classification exists among the samples, performing a fourth step of weighting parameter Mahalanobis distance classification. And fifthly, finally judging the hardware Trojan horse according to the significance level. According to the method, the sample can be weighted and the characteristic parameters of the sample can be adjusted under the condition that the discrimination and classification of the hardware Trojan horse are invalid through the conventional Mahalanobis distance, so that the difference of tiny characteristics in the sample to be detected is further amplified, the discrimination and classification of the Trojan horse sample and the Trojan horse-free sample are finally achieved, and the hardware Trojan horse detection rate is effectively improved.

Description

Hardware Trojan horse detection method based on weighting parameter Mahalanobis distance
Technical Field
The invention relates to a hardware Trojan detection method of an integrated circuit, in particular to a hardware Trojan detection method based on a principal component analysis method, and belongs to the field of integrated circuit safety.
Background
The hardware trojan refers to a circuit which is already processed or a malicious circuit which is implanted in an original circuit in the design or manufacturing process of an integrated circuit, and can complete effective attack in the forms of integrated circuit failure, information leakage, denial of service and the like under the specific conditions set by an attacker. As the design of integrated circuits becomes more sophisticated, the separation of the design and production process flow thereof causes the intrusion risk of hardware trojans to increase greatly.
The existing hardware Trojan horse detection technology comprises the following steps: reverse engineering, logic function testing, built-in testing and side channel analysis. The side channel analysis technology has the advantages of no damage, low detection cost, high detection efficiency and the like, and becomes a hot spot for the research of the hardware Trojan detection method in recent years. When the overall scale of the integrated circuit to be tested is large and the area occupied by the implanted hardware Trojan is small, the change of a side channel signal is small, but due to the influence of process deviation and environmental noise, the conventional Mahalanobis distance judgment method has an unsatisfactory judgment effect or fails on the hardware Trojan with a small occupation ratio.
The invention provides a hardware Trojan horse detection method based on a weighting parameter Mahalanobis distance. The method adopts the weight to carry out weighting in the discriminant and the power consumption matrix to carry out double-parameter adjustment to judge whether the hardware Trojan horse exists or not, and can be applied to the hardware Trojan horse discrimination with smaller area occupation ratio.
Disclosure of Invention
1. The purpose is as follows: aiming at the defect that the Mahalanobis distance discrimination method in the existing side channel analysis technology processes data, the method provided by the invention utilizes the weights of all dimensional characteristics of a sample to carry out weighting in a discrimination formula and carry out double-parameter adjustment on a generated power consumption matrix under the failure condition of the conventional Mahalanobis distance discrimination method. And finally, comparing and judging the Mahalanobis distance value of the sample to be detected and the confidence interval of the standard sample, wherein the method has higher hardware Trojan detection rate than the conventional Mahalanobis distance method under the condition of the same sample group number.
2. The technical scheme is as follows:
the method of the invention comprises 5 steps:
and in the first process, sample data of the ports of the standard FPGA and the FPGA to be tested is collected.
And step two, classifying the Mahalanobis distance between the standard and the sample to be detected.
And thirdly, selecting a significant level and judging a hardware Trojan horse.
And fourthly, weighting the parameter Mahalanobis distance classification.
And fifthly, finally judging the hardware trojan.
The first process specifically comprises the following steps:
the method comprises the steps that a direct-current stabilized voltage supply is turned on to supply power to an FPGA and a PCB detection board, the known standard FPGA is a hardware-free Trojan sample, one of two FPGAs to be detected is implanted with a hardware Trojan sample A with the area ratio of 0.3%, the other is a hardware-free Trojan sample B, and the three FPGAs are classified into the same type series of FPGAs. The high-performance oscilloscope and the signal probe acquire currents of a standard FPGA and a FPGA to be detected under normal operation through measuring and sampling pins of a PCB (printed circuit board), so that currents of p ports are obtained, and n groups of power consumption data are obtained by repeating the operations for n times.
In the second process, the mahalanobis distance classification process of the sample is shown in fig. 3, and specifically includes:
step (1) generates a power consumption matrix X, Y, Z of n × p size for each block of the obtained n sets of power consumption data.
Step (2) calculates a standard sample matrix and generates a determinant of the mean of 1 × p.
Figure BDA0002700996980000021
And (3) solving covariance matrixes of the three samples and obtaining by inversion:
Figure BDA0002700996980000031
Figure BDA0002700996980000032
Figure BDA0002700996980000033
step (4)
The sample average matrix n u1 generated from the average u1 is subtracted from the three sample matrices to generate a difference matrix D1、D2、D3。
Figure BDA0002700996980000034
Figure BDA0002700996980000035
Figure BDA0002700996980000036
Transpose the generated matrix to E1、E2And E3。
E1＝D1 T E2＝D2 T E3＝D3 T
Step (5) c1、c2And c3The format is 1 × n determinant. And putting the calculated result in an established Matlab GUI table and generating a conventional Mahalanobis distance scatter diagram of the standard sample and the sample to be detected by using a coordinate axis frame module.
Figure BDA0002700996980000037
c1Is a conventional mahalanobis distance value of a standard sample, c2Is the conventional mahalanobis distance value, c, of the sample A to be measured3Is the conventional mahalanobis distance value of the sample B to be measured,
the third process specifically comprises the following steps:
step (6) calculating a confidence interval of the Mahalanobis distance of the training sample by using interval estimation, wherein the distribution of the process deviation is random and the Mahalanobis distance does not change the distribution condition; the uniformly distributed confidence interval calculation formula is as follows:
UL＝M+V×ST
LL＝M-V×ST
wherein UL is the upper confidence limit, and LL is the lower confidence limit; m is the Mahalanobis distance mean value of the training sample, and ST is the Mahalanobis distance standard deviation of the training sample; the value of V is related to the confidence, and when the confidence is 95%, V is 1.96.
Determining the confidence interval of the standard sample, DbFor the length of the standard confidence interval, the distance value of the sample to be measured is compared with the confidence interval of the standard sample, DdThe number of the samples to be detected in the confidence interval is shown.
Step (7) if Dd<40% n, calculating the hardware Trojan horse detection rate of the sample to be detected, wherein the detection rate is
Figure BDA0002700996980000041
If D isd>40% n, then enter process four stage.
The process four, the mahalanobis distance classification flow of the sample weighting parameters is shown in fig. 4;
the method specifically comprises the following steps:
step (8) calculates a standard sample matrix and generates a determinant of the mean of 1 × p.
Figure BDA0002700996980000042
And (9) establishing dual parameters r and k of the standard sample and the sample matrix to be detected. Using Matlab [ a, b ]]Ndgrid (1: N) and a ═ a (: b (:)]Generating two 1-N parameter full permutations, N being selected according to a power consumption sample judgment pre-experiment for judgment, M3Size (a,1) is the number of rows of the parameter combination, r is a (i,1) th row 1 th value, and k is a (i,2) th row 2 nd value.
And (10) uniformly adjusting double parameters of the standard sample matrix and the sample matrix to be detected. Respectively carrying out 1-M operations on the M3 group data of the combination parameters3And (6) circulating. The three sample matrixes are subjected to parameter adjustment to obtain the subtraction of X1, Y1 and Z1The sample mean matrix n x u1 generated from the mean u1 is removed to generate a difference matrix D4、D5、D6。
Figure BDA0002700996980000043
Figure BDA0002700996980000051
Figure BDA0002700996980000052
Figure BDA0002700996980000053
Figure BDA0002700996980000054
Figure BDA0002700996980000055
Transpose the generated matrix to E4、E5And E6。
E4＝D4 T E5＝D5 T E6＝D6 T
Solving the covariance matrix of three samples X1, Y1 and Z1 and inverting to obtain:
Figure BDA0002700996980000056
Figure BDA0002700996980000057
Figure BDA0002700996980000058
and (11) determining the weights of the standard sample and the sample to be measured. And randomly selecting p groups of samples from the standard sample X1 to form a matrix X2, and randomly selecting p groups of samples from the samples Y1 and Z1 to be detected to form matrices Y2 and Z2.
Step (12) finds a eigenvector matrix with a1 being Y2, b1 being Y2 eigenvalue matrix, and finds the column m corresponding to the largest eigenvalue in b1, w1 being a1(: m1)/sum (a1(: m1)) being the weight of the standard sample, by using the Matlab program [ a1, b1] ═ eig (Y2),
and (13) repeating the step (21) on the matrixes Y2 and Z2 to obtain the weights of the w2 and the w3 to-be-detected samples. Q1、Q2、Q3A diagonal matrix formed by components in a weight vector of the p-dimensional feature, wherein
Figure BDA0002700996980000061
And finally, taking the square value of the square of the values of w1, w2 and w3 as a weight matrix of a diagonal line.
Figure BDA0002700996980000062
Figure BDA0002700996980000063
Figure BDA0002700996980000064
And (14) putting the output results c3, c4 and c5 in the established uitable4 and generating a Mahalanobis distance discrimination scatter diagram of the weighting parameters of the standard sample and the sample to be detected by using a coordinate axis frame module.
Figure BDA0002700996980000065
Figure BDA0002700996980000066
c4Is a weighted parameter of the standard sample, mahalanobis distance value, c5Is the weighted parameter Mahalanobis distance value, c of the sample A to be measured6And the weighted parameter is the mahalanobis distance value of the sample B to be detected.
The fifth process specifically comprises the following steps:
step (15) of discriminating 1-M between Mahalanobis distances as weighting parameters each time3After circulation, comparing the distance value of the sample to be measured with the confidence interval of the standard sample, DbTo the length of the standard confidence interval, DdThe number of the samples to be detected in the confidence interval is shown. If D isd<40% n, calculating the hardware Trojan horse detection rate of the sample to be detected, wherein the detection rate is
Figure BDA0002700996980000067
If D is obtained in the comparison after the weighted parameter Mahalanobis distance discrimination loop program is operated each timed>40% n, then M is completed3And the Mahalanobis distance discrimination circulation program of the secondary weighting parameter obtains the hardware-free Trojan horse of the sample to be detected.
3. The advantages and the effects are as follows:
1) the method can further divide the sample characteristics to realize more precise classification by combining weighting and double parameters under the condition that the discrimination and classification of the samples to be detected by the conventional Mahalanobis distance discrimination are invalid.
2) The method can be effectively verified in the aspect of improving the hardware Trojan horse detection rate.
2) The invention enables the data processing efficiency and the standardization of hardware Trojan horse detection based on Matlab GUI.
Drawings
FIG. 1 is a flow chart of a hardware Trojan horse detection method based on a weighting parameter Mahalanobis distance;
FIG. 2 is a schematic view of the present invention;
FIG. 3 is a flow chart of a sample conventional Mahalanobis distance classification;
FIG. 4 is a flowchart of a Mahalanobis distance classification process for sample weighting parameters;
FIG. 5 is a Matlab GUI conventional Mahalanobis distance discrimination scatter plot;
fig. 6 is a Matlab GUI weighting parameter mahalanobis distance discrimination scattergram.
The symbol numbers in the figures are illustrated as follows:
n: number of sample groups.
FIG. 2:
r: adjusting parameters of a sample matrix;
k: adjusting parameters of a sample matrix;
Dd: the number of the samples to be detected in the confidence interval is calculated;
Db: is the standard confidence interval length;
i: adjusting the parameter cycle times;
m: adjusting the maximum combination number of the parameters;
FIG. 3:
c: a sample conventional mahalanobis distance value;
sqrt: squaring;
d: a difference matrix of the standard sample matrix average value subtracted from the sample matrix to be measured;
s: an inverse of the sample covariance matrix;
e: and (5) transposing the difference matrix.
FIG. 4:
w: sample weight;
c: a sample weighting parameter mahalanobis distance value;
sqrt: squaring;
d: the difference matrix after parameter adjustment;
s: the inverse matrix of the sample covariance matrix after parameter adjustment;
e: transposing the difference matrix after parameter adjustment;
q: a weight matrix.
Detailed Description
1) The method comprises the steps that 1 standard FPGA and 2 FPGAs to be tested are known, one of the 2 FPGAs to be tested is implanted with a time sequence type hardware Trojan, and the other is not implanted with the hardware Trojan. Under the condition that the DC stabilized voltage supply supplies power to the FPGA and the PCB detection board, a high-performance oscilloscope and a signal probe are utilized to detect the current of the normal operation of the FPGA by measuring the voltage difference between two ends of a sampling resistor for collection, thereby obtaining the current of 26 ports, repeatedly collecting 100 times within 30s to generate a power consumption matrix of 100 multiplied by 26,
2) a power consumption matrix X, Y, Z of size 100 × 26 is generated for each of the 100 sets of power consumption data obtained for each block, respectively.
Figure BDA0002700996980000081
Figure BDA0002700996980000091
Figure BDA0002700996980000092
A standard sample matrix is calculated and a determinant of the mean of 1 × 26 is generated.
u1＝[34.89 0.15 … 1.02 30.40]
Step (2.1) of solving the covariance matrixes of the three samples and obtaining S by inversion1、S2、S3：
Figure BDA0002700996980000093
Figure BDA0002700996980000094
Figure BDA0002700996980000095
Step (2.2) subtracts the sample mean matrix n × u1 generated from the mean u1 for the three sample matrices, generating a difference matrix D1、D2、D3。
Figure BDA0002700996980000096
Figure BDA0002700996980000097
Figure BDA0002700996980000101
Step (2.3) transposes the generated matrix into E1、E2And E3。
E1＝D1 T E2＝D2 T E3＝D3 T
Step (2.4) c1、c2And c3Each 1 × 100 determinant. And putting the calculated result in an established Matlab GUI table, and generating a conventional Mahalanobis distance scatter diagram of the standard sample and the sample to be detected by using a coordinate axis frame module, as shown in FIG. 5.
c1＝[4.4836 5.1461 … 4.6042 5.3685]
c2＝[4.8152 5.7202 … 6.7900 5.6635]
c3＝[4.9153 5.1223 … 4.8601 5.7636]
c1Is a conventional mahalanobis distance value of a standard sample, c2Is the conventional mahalanobis distance value, c, of the sample A to be measured3Is the conventional mahalanobis distance value of the sample B to be measured,
3) determining the confidence interval of the standard sample, DbIs the length of the standard confidence interval [4.21,6.02 ]]Comparing the distance value of the sample to be measured with the confidence interval of the standard sample, DdThe number of the samples to be detected in the confidence interval is 82 and 85, and D is satisfiedd>40% n, step 4.
4) The process of classifying the mahalanobis distance as a sample weighting parameter is shown in fig. 4, which specifically includes:
step (4.1) calculates the standard sample matrix and generates a determinant of the mean of 1 × 26.
u1＝[34.89 0.15 … 1.02 30.40]
Step (4.2) solving covariance matrixes of the three samples of X1, X2 and X3 after parameter adjustment and obtaining S through inversion4、S5、S6：
Figure BDA0002700996980000102
Figure BDA0002700996980000111
Figure BDA0002700996980000112
And (4.3) determining the weights of the standard sample and the sample to be measured. And randomly selecting p groups of samples from the standard sample X1 to form a matrix X2, and randomly selecting p groups of samples from the samples Y1 and Z1 to be detected to form matrices Y2 and Z2. By using Matlab program [ a1, b1] ═ eig (Y2), eigenvector matrix with a1 being Y2 and b1 being Y2 eigenvalue matrix were obtained, and m in column b1 corresponding to the largest eigenvalue was found, and w1 ═ a1(: m1)/sum (a1(: m1)) was used as the weight of the standard sample.
And (4.4) repeating the step (16.2) on the matrixes Y2 and Z2 to obtain the weights of the w2 and w3 to-be-detected samples. Q1、Q2、Q3A diagonal matrix formed by components in a weight vector of the p-dimensional feature, wherein
Figure BDA0002700996980000113
And finally, taking the square value of the square of the values of w1, w2 and w3 as a weight matrix of a diagonal line.
Figure BDA0002700996980000114
Figure BDA0002700996980000115
Figure BDA0002700996980000116
And (4.5) establishing dual parameters r and k of the standard sample and the sample matrix to be detected. Using Matlab [ a, b ]]Ndgrid (1:5) and a ═ a (: b (:)]Generating two full permutations of parameters, M, of 1-103The parameter combination row number is 25, r, a (i,1) is the ith row number 1, and k, a (i,2) is the ith row number 2.
And (4.6) performing systematic adjustment on double parameters of the standard sample matrix and the sample matrix to be detected. And (3) performing combination circulation on 100 groups of data of the combination parameters, wherein r is 1-10, and k is 1-10. The sample average matrix 100 × u1 generated from the average u1 is subtracted from the three sample matrices to generate a difference matrix D4、D5、D6。
Figure BDA0002700996980000121
Figure BDA0002700996980000122
Figure BDA0002700996980000123
Step (4.7) transposes the generated matrix into E4、E5And E6。
E4＝D4 T E5＝D5 T E6＝D6 T
And (4.8) putting the output results c3, c4 and c5 in the established uitable4 and generating a conventional mahalanobis distance scatter diagram of the standard sample and the sample to be detected by using a coordinate axis frame module, as shown in fig. 6.
c1＝[4.193 4.359 … 4.1697 4.0370]
c2＝[14.473 8.803 … 28.131 11.835]
c3＝[5.0245 6.280 … 3.488 6.05]
c4Is a weighted parameter of the standard sample, mahalanobis distance value, c5Is the weighted parameter Mahalanobis distance value, c of the sample A to be measured6And the weighted parameter is the mahalanobis distance value of the sample B to be detected.
5) Judging the calculated weighting parameter Mahalanobis distance value, specifically:
and (5.1) judging once after 1-100 cycles of judging the Mahalanobis distance of the weighting parameter every time. When the loop to the parameter value r 10, k 3 is performed, the standard sample confidence interval is [6.286,3.151 ]]The number of the confidence intervals in the standard sample in the sample A to be detected is 90, and the number of the confidence intervals in the sample B to be detected is 12. Judging that the sample A to be detected contains the hardware Trojan horse, and calculating the hardware Trojan horse detection rate of the sample A
Figure BDA0002700996980000124
Sample B has no hardware trojan.

Claims (6)

1. A hardware Trojan horse detection method based on a weighting parameter Mahalanobis distance is characterized by comprising the following steps:
in the first process, sample power consumption data of ports of a standard FPGA and an FPGA to be tested are collected.
And step two, classifying the standard and the sample to be detected by the conventional Mahalanobis distance.
And thirdly, selecting the significance level and judging the hardware Trojan horse.
And fourthly, weighting the parameter Mahalanobis distance classification.
And fifthly, finally judging the hardware trojan.
2. The hardware Trojan horse detection method based on the Mahalanobis distance as the weighting parameter of claim 1, wherein the first process comprises a Matlab GUI interface, a DC stabilized power supply, a PCB detection board and PFGA, a high-performance oscilloscope and a signal probe.
3. The hardware Trojan horse detection method based on the Mahalanobis distance as claimed in claim 1, wherein the second process comprises the determinant u1 of the average number of the standard sample matrix 1 × p, the inverse matrix of the covariance matrix of three samples, the subtraction of the average value u1 of the standard sample from the n groups of data of two sample matrices by 1-n cycles and the generation of the transpose matrix, and finally the performance of the difference matrix, the transpose matrix and the inverse matrix is squared. And finally, generating a conventional mahalanobis distance scatter diagram by using conventional mahalanobis distance values of the three samples and adopting a Matlab GUI.
4. The method as claimed in claim 1, wherein the third process comprises determining whether the hardware Trojan horse is contained in the sample to be tested. Firstly, selecting a sample significance level, if a judgment condition is met, calculating the hardware Trojan horse detection rate, and if the judgment condition is not met, executing a process four.
5. The hardware Trojan horse detection method based on the weighting parameter Mahalanobis distance as claimed in claim 1, wherein the process IV comprises the steps of calculating the weight of the characteristic dimensions of the standard sample and the sample to be detected, and performing the weighting and the two-parameter cyclic adjustment of the two sample power consumption matrixes in the conventional Mahalanobis distance operation mode.
6. The method as claimed in claim 1, wherein the process five includes determining whether the hardware trojan is contained in the sample to be tested. And analyzing the generated discrimination result and the Martensis distance scatter diagram of the sample. And if the threshold condition is met, calculating the hardware Trojan horse detection rate of the sample to be detected, and if the threshold condition is not met, determining that the sample to be detected does not have the hardware Trojan horse.
CN202011028757.3A 2020-09-25 2020-09-25 Hardware Trojan horse detection method based on weighting parameter Mahalanobis distance Pending CN112163245A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011028757.3A CN112163245A (en) 2020-09-25 2020-09-25 Hardware Trojan horse detection method based on weighting parameter Mahalanobis distance

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011028757.3A CN112163245A (en) 2020-09-25 2020-09-25 Hardware Trojan horse detection method based on weighting parameter Mahalanobis distance

Publications (1)

Publication Number Publication Date
CN112163245A true CN112163245A (en) 2021-01-01

Family

ID=73862922

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011028757.3A Pending CN112163245A (en) 2020-09-25 2020-09-25 Hardware Trojan horse detection method based on weighting parameter Mahalanobis distance

Country Status (1)

Country Link
CN (1) CN112163245A (en)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020156549A1 (en) * 2001-04-20 2002-10-24 Shunji Hayashi Controlling method for manufacturing process
CN103488941A (en) * 2013-09-18 2014-01-01 工业和信息化部电子第五研究所 Hardware Trojan horse detection method and hardware Trojan horse detection system
CN105184661A (en) * 2015-07-14 2015-12-23 国家电网公司 Grid monitoring signal analysis method based on weighted Mahalanobis distance discrimination

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020156549A1 (en) * 2001-04-20 2002-10-24 Shunji Hayashi Controlling method for manufacturing process
CN103488941A (en) * 2013-09-18 2014-01-01 工业和信息化部电子第五研究所 Hardware Trojan horse detection method and hardware Trojan horse detection system
CN105184661A (en) * 2015-07-14 2015-12-23 国家电网公司 Grid monitoring signal analysis method based on weighted Mahalanobis distance discrimination

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
朱惠倩等: "聚类分析的一种改进方法", 《湖南文理学院学报》 *
王建新等: "基于改进欧式距离的硬件木马检测", 《计算机工程》 *

Similar Documents

Publication Publication Date Title
US10539613B2 (en) Analog circuit fault diagnosis method using single testable node
CN109729091A (en) A kind of LDoS attack detection method based on multiple features fusion and CNN algorithm
CN110475246B (en) Malicious anchor node detection method based on isolated forest and sequential probability ratio detection
CN107733541A (en) Method, apparatus, equipment and the computer-readable recording medium of frequency spectrum perception
CN103150733A (en) Self-adapting multi-stage weighted median filtering algorithm applied to digital images
CN107480561A (en) Hardware Trojan horse detection method based on few state node traverses
CN114062850B (en) Double-threshold power grid early fault detection method
CN107360127A (en) A kind of Denial of Service attack detection method at a slow speed based on AEWMA algorithms
CN111523588A (en) Method for classifying APT attack malicious software traffic based on improved LSTM
CN115032606B (en) Constant false alarm detector based on local minimum selected unit average
CN114089325A (en) Extended target detection method and system when interference information is uncertain
CN111797887A (en) Anti-electricity-stealing early warning method and system based on density screening and K-means clustering
CN104880696B (en) One kind is based on the preferable multi-channel target detection method of passage
CN111600877A (en) LDoS attack detection method based on MF-Ada algorithm
CN110008987B (en) Method and device for testing robustness of classifier, terminal and storage medium
CN112163245A (en) Hardware Trojan horse detection method based on weighting parameter Mahalanobis distance
CN110298200B (en) ASIC chip hardware back door detection method based on temperature statistics feature analysis
CN110086860B (en) Data anomaly detection method and device under Internet of things big data environment
Biswas et al. Biological averaging in RNA-seq
CN105046171B (en) A kind of hardware Trojan horse detection method based on addition of waveforms
CN114089307B (en) Radar detection and classification method and system under target and interference conditions
Wang et al. Anomaly detection and array diagnosis in wireless networks with multiple antennas: Framework, challenges and tools
CN113014361B (en) BPSK signal confidence test method based on graph
CN109684881B (en) Bypass detection method and device
CN110298204B (en) ASIC chip hardware Trojan diagnosis method based on temperature field time-space effect

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20210101

WD01 Invention patent application deemed withdrawn after publication