CN103198251B - Hardware Trojan horse recognition method based on neural network - Google Patents

Hardware Trojan horse recognition method based on neural network Download PDF

Info

Publication number
CN103198251B
CN103198251B CN201310103424.6A CN201310103424A CN103198251B CN 103198251 B CN103198251 B CN 103198251B CN 201310103424 A CN201310103424 A CN 201310103424A CN 103198251 B CN103198251 B CN 103198251B
Authority
CN
China
Prior art keywords
chip
channel information
side channel
matrix
trojan horse
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.)
Expired - Fee Related
Application number
CN201310103424.6A
Other languages
Chinese (zh)
Other versions
CN103198251A (en
Inventor
王晨旭
罗敏
喻明艳
王进祥
李�杰
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.)
Harbin Institute of Technology
Harbin Institute of Technology Weihai
Original Assignee
Harbin Institute of Technology Weihai
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 Harbin Institute of Technology Weihai filed Critical Harbin Institute of Technology Weihai
Priority to CN201310103424.6A priority Critical patent/CN103198251B/en
Publication of CN103198251A publication Critical patent/CN103198251A/en
Application granted granted Critical
Publication of CN103198251B publication Critical patent/CN103198251B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Landscapes

  • Storage Device Security (AREA)
  • Image Analysis (AREA)

Abstract

The invention provides a hardware Trojan horse recognition method based on a neural network. The problem that an existing recognition method does not need manual observation and is low in efficiency is solved, and recognition intelligentization of a hardware Trojan horse chip is achieved. The method comprises the following steps of first obtaining side channel information of a chip to be detected and performing data pre-processing; selecting a part of the chip to be detected to perform back subdivision analysis and confirming whether a back subdivision chip contains hardware Trojan horse or not; utilizing the back subdivision chip which does not contain the hardware Trojan horse to build a chip characteristic space through pro-processed side channel information; projecting all of chips to be detected into the characteristic space through a pro-processed side channel information matrix to obtain a side channel information characteristic data matrix; utilizing side channel information characteristic data of the back subdivision chip and a corresponding target output value to build and train the neural network; delivering the side channel information characteristic data of a tested chip to the trained neural network to be distinguished and output and achieving hardware Trojan horse recognition.

Description

A kind of hardware Trojan horse recognition method based on neural network
Technical field
The present invention relates to the detection and Identification field of hardware Trojan horse chip, be specifically related to a kind of hardware Trojan horse recognition method based on neural network.
Background technology
Along with being separated of each links such as integrated circuit (IC) design, manufacture, test, encapsulation, the chip that user uses is become large by the possibility that ax-grinder implants hardware Trojan Horse circuit, brings serious threat to the reliability of information security field and chip.At present, for the detection and indentification of hardware Trojan horse chip, what in most cases adopt is extract chip side channel information characteristics value, to set up the technology of the side channel information fingerprint base of chip.The general step of this technology is, first from a collection of chip some chips of random selecting as sample; Detect the side channel information of sample chip, comprise power consumption, the side such as electromagnetism or heat channel information; Counter cuing open is destroyed to sample chip, obtains chip type, determine that whether chip is containing hardware Trojan horse; Utilize the method for Eigenvalues Decomposition to extract sample chip side channel information eigenwert, set up the side channel finger print information of chip; Other chips are directly tested to their side channel information, and extract the eigenwert of side channel information, with fingerprint base comparison, draw their chip type.For Eigenvalue Extraction Method, general Karhunen-Loeve transformation, do not process containing the side channel information data of the sample chip of hardware Trojan horse determining, set up the sample characteristics subspace of non-wooden horse chip, the side channel information of non-wooden horse chip is projected to this subspace, make its projected data image, and set up fingerprint base; With batch same data processing method of other chips, side channel information is projected to this subspace, obtains respective side channel characteristic information, compare with the fingerprint base information obtained, judge the whether implanted hardware Trojan horse of chip.This technology needs to draw, and needs manually the image of test data and the image of finger print information to be compared, and be easy to bring the collimation error, recognition efficiency is low.In addition, because wooden horse type is different, the side channel information of the chip infecting wooden horse is not quite similar, increases the difficulty by image recognition hardware Trojan horse further.
Summary of the invention
The present invention comes whether adopt manual observation image containing needing in the method for hardware Trojan horse in identification chip to solve the existing chip side channel information that utilizes, there is large, the inefficient problem of error, provide a kind of intellectuality, the hardware Trojan horse recognition method based on neural network that accuracy rate is high.
Hardware Trojan horse recognition method based on neural network of the present invention comprises the following steps:
Step one, for all chips to be detected, data prediction is carried out to its side channel information, obtain the side channel information matrix of pretreated all chip to be detected;
Step 2, the part choosing chip to be detected cut open chip as counter, and other remains chip to be detected and is called test chip, oppositely analyzes the described anti-chip that cuts open, and determine that whether each anti-chip that cuts open is containing hardware Trojan horse;
Step 3, take out from the side channel information matrix described in step one through step 2 determine not containing the anti-side channel information cutd open corresponding to chip of hardware Trojan horse, utilize this side channel information to set up the feature space of chip;
Step 4, by feature space described in the side channel information matrix projection to step 3 of chip to be detected after pretreatment, obtain the side channel information characteristic matrix of all chips to be detected;
Step 5, take out from the side channel information characteristic matrix obtained after step 4 process and instead cut open the training sample of the characteristic corresponding to chip as neural network, utilize described training sample and its corresponding target output value to set up and neural network training;
Step 6: take out the side channel information characteristic corresponding to test chip from the side channel information characteristic matrix obtained after step 4 process, is carried out differentiation in the neural network that described data feeding step 5 has been trained to export, whether distinguish corresponding test chip containing hardware Trojan horse according to differentiation output valve.
Advantage of the present invention: a kind of maximum feature of the hardware Trojan horse recognition method based on neural network is the side channel characteristics data using known type chip, set up and neural network training, make the neural network after training can the side channel characteristics data of automatic distinguishing wooden horse chip and the side channel characteristics data of non-wooden horse chip, reach the intelligent object of hardware Trojan horse chip identification thus, improve recognition efficiency and the accuracy rate of hardware Trojan horse, overcome the defect of existing recognition methods.
Embodiment
A kind of hardware Trojan horse recognition method based on neural network described in present embodiment, the method comprises the following steps:
Step one, for all chips to be detected, data prediction is carried out to its side channel information, obtain the side channel information matrix of pretreated all chip to be detected;
Step 2, the part choosing chip to be detected cut open chip as counter, and other remains chip to be detected and is called test chip, oppositely analyzes the described anti-chip that cuts open, and determine that whether each anti-chip that cuts open is containing hardware Trojan horse;
Step 3, take out from the side channel information matrix described in step one through step 2 determine not containing the anti-side channel information cutd open corresponding to chip of hardware Trojan horse, utilize this side channel information to set up the feature space of chip;
Step 4, by feature space described in the side channel information matrix projection to step 3 of chip to be detected after pretreatment, obtain the side channel information characteristic matrix of all chips to be detected;
Step 5, take out from the side channel information characteristic matrix obtained after step 4 process and instead cut open the training sample of the characteristic corresponding to chip as neural network, utilize described training sample and its corresponding target output value to set up and neural network training;
Step 6: take out the side channel information characteristic corresponding to test chip from the side channel information characteristic matrix obtained after step 4 process, is carried out differentiation in the neural network that described data feeding step 5 has been trained to export, whether distinguish corresponding test chip containing hardware Trojan horse according to differentiation output valve.
The process that the side channel information treating detection chip in step one carries out data prediction is:
Step is sampled one by one, to the side channel information of time dependent chip to be detected, obtains side channel information matrix , represent the number of chip to be detected, represent the sampling number of side channel information;
Step one two, offside channel information matrix carry out centralization process by row, obtain pretreated side channel information matrix according to the following formula ,
Wherein, represent pretreated side channel information matrix ioK jthe value of row, represent side channel information matrix before pre-service the ioK jthe value of row, represent side channel information matrix before pre-service the jthe mean value of column data.
The process setting up the feature space of chip through pretreated side channel information matrix not cuing open chip containing hardware Trojan horse counter is utilized to be in step 3:
Step 3 one, be not expressed as matrix containing the anti-side channel information cutd open corresponding to chip of hardware Trojan horse by described in step 3 form, wherein it is one ndimensional vector, represents the ithe different value of moment different chip side channel information, and nrepresent and do not cut open chip number containing the counter of hardware Trojan horse, represent the sampling number of side channel information;
Step 3 two, ask matrix covariance matrix , of covariance matrix ioK jthe element representation of row is
Wherein, with represent column vector respectively with average, nmeaning and step 3 one in ndefinition identical, E is mathematical expectation, then matrix covariance matrix for:
；
Step 3 three, utilize equation , ask covariance matrix eigenwert and with corresponding proper vector , and constitutive characteristic vector matrix ;
Step 3 four, by T eigenwert by from big to small order arrangement, matrix described in step 3 three in each column vector corresponding to each eigenwert be also adjusted accordingly according to the adjustment order of eigenwert and obtain new matrix ;
Step 3 five, to matrix described in step 3 four carry out unit orthogonalization and obtain matrix , by the determined space of each column vector be the chip features space that will obtain.
Be objective chip choosing DES cryptographic algorithm chip, choosing SMIC18 is target process, process noise is under the prerequisite of 5%, when chip area is greater than 7 ‰ compared with objective chip total area shared by hardware Trojan horse circuit, adopts the method for the invention can reach the recognition accuracy of 100%.

Claims (3)

1. based on a hardware Trojan horse recognition method for neural network, it is characterized in that, comprise the following steps:
Step one, for all chips to be detected, data prediction is carried out to its side channel information, obtain the side channel information matrix of pretreated all chip to be detected;
Step 2, the part choosing chip to be detected cut open chip as counter, and other remains chip to be detected and is called test chip, oppositely analyzes the described anti-chip that cuts open, and determine that whether each anti-chip that cuts open is containing hardware Trojan horse;
Step 3, take out from the side channel information matrix described in step one through step 2 determine not containing the anti-side channel information cutd open corresponding to chip of hardware Trojan horse, utilize this side channel information to set up the feature space of chip;
Step 4, by feature space described in the side channel information matrix projection to step 3 of chip to be detected after pretreatment, obtain the side channel information characteristic matrix of all chips to be detected;
Step 5, take out from the side channel information characteristic matrix obtained after step 4 process and instead cut open the training sample of the characteristic corresponding to chip as neural network, wherein, contain the positive class sample not containing hardware Trojan horse and the negative class sample containing hardware Trojan horse, utilize two class target output values of this two classes sample and its correspondence to set up and neural network training
Step 6, from the side channel information characteristic matrix obtained after step 4 process, take out side channel information characteristic corresponding to test chip, is carried out differentiation in the neural network that described data feeding step 5 has been trained to export, whether distinguish corresponding test chip containing hardware Trojan horse according to differentiation output valve.
2. according to claim 1 based on the hardware Trojan horse recognition method of neural network, it is characterized in that, the process that the side channel information treating detection chip in step one carries out data prediction is:
Step is sampled one by one, to the side channel information of time dependent chip to be detected, obtains side channel information matrix I n × T, N represents the number of chip to be detected, and T represents the sampling number of side channel information;
Step one two, offside channel information matrix carry out centralization process by row, obtain pretreated side channel information matrix I_cent according to the following formula n × T,
I_cent ij＝I ij-I_mean j
I _ mean j = 1 N Σ i = 1 N I ij
Wherein, I_cent ijrepresent the value that pretreated side channel information matrix i-th row j arranges, I ijrepresent the value that the side channel information matrix i-th row j before pre-service arranges, I_mean jrepresent the mean value of the side channel information matrix jth column data before pre-service.
3. according to claim 1 based on the hardware Trojan horse recognition method of neural network, it is characterized in that, in step 3, utilize the process setting up the feature space of chip through pretreated side channel information matrix not cuing open chip containing hardware Trojan horse counter to be:
Step 3 one, be not expressed as matrix X=[X containing the anti-side channel information cutd open corresponding to chip of hardware Trojan horse by described in step 3 1, X 2..., X t] form, wherein X ia n dimensional vector, i=1,2 ..., T, represent the different value of the i-th moment different chip side channel information, and n represents that, not containing the anti-chip number of cuing open of hardware Trojan horse, T represents the sampling number of side channel information;
Step 3 two, ask the covariance matrix C of matrix X f, the element representation that the i-th row j of covariance matrix arranges is
Σ ij = cov ( X i , X j ) = E [ ( X i - μ i ) · ( X j - μ j ) ] = 1 n - 1 Σ m = 1 n ( X mi - μ i ) ( X mj - μ j )
Wherein, μ iand μ jrepresent column vector X respectively iand X javerage, the meaning of n is identical with the definition of n in step 3 one, and E is mathematical expectation, then the covariance matrix C of matrix X ffor:
Step 3 three, utilize equation C fΦ i=λ iΦ i, i=1,2 ..., T, asks covariance matrix C feigenvalue λ iand with λ icorresponding proper vector Φ i, and constitutive characteristic vector matrix Φ=[Φ 1, Φ 2..., Φ t];
Step 3 four, by T eigenwert by order arrangement from big to small, each column vector corresponding to each eigenwert in matrix Φ described in step 3 three is also adjusted accordingly obtains new matrix Φ ' according to the adjustment order of eigenwert;
Step 3 five, unit orthogonalization is carried out to matrix Φ ' described in step 3 four obtain matrix A, be the chip features space that will obtain by the determined space of each column vector of A.
CN201310103424.6A 2013-03-28 2013-03-28 Hardware Trojan horse recognition method based on neural network Expired - Fee Related CN103198251B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201310103424.6A CN103198251B (en) 2013-03-28 2013-03-28 Hardware Trojan horse recognition method based on neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201310103424.6A CN103198251B (en) 2013-03-28 2013-03-28 Hardware Trojan horse recognition method based on neural network

Publications (2)

Publication Number Publication Date
CN103198251A CN103198251A (en) 2013-07-10
CN103198251B true CN103198251B (en) 2015-07-08

Family

ID=48720800

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201310103424.6A Expired - Fee Related CN103198251B (en) 2013-03-28 2013-03-28 Hardware Trojan horse recognition method based on neural network

Country Status (1)

Country Link
CN (1) CN103198251B (en)

Families Citing this family (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103532957B (en) * 2013-10-18 2017-09-15 电子科技大学 A kind of long-range shell behavioral values device and method of wooden horse
CN103698687B (en) * 2013-12-18 2017-01-04 工业和信息化部电子第五研究所 In integrated circuit hardware Trojan horse detection signal processing method and system
CN104215894B (en) * 2014-08-28 2017-04-05 工业和信息化部电子第五研究所 IC Hardware Trojan detecting method and system
CN104316861B (en) * 2014-10-16 2017-05-10 工业和信息化部电子第五研究所 integrated circuit hardware Trojan detection method and system
CN104330721B (en) * 2014-10-29 2017-03-08 工业和信息化部电子第五研究所 IC Hardware Trojan detecting method and system
CN105893876A (en) * 2016-03-28 2016-08-24 工业和信息化部电子第五研究所 Chip hardware Trojan horse detection method and system
CN106845287A (en) * 2017-01-25 2017-06-13 天津大学 Hardware Trojan horse detection method based on multi-parameter correlation
CN107703186A (en) * 2017-09-26 2018-02-16 电子科技大学 Hardware Trojan horse detection method based on chip temperature field-effect
CN108052840A (en) * 2017-11-13 2018-05-18 天津大学 Hardware Trojan horse detection method based on neutral net
CN108154051A (en) * 2017-11-23 2018-06-12 天津科技大学 A kind of hardware Trojan horse detection method of discrimination based on support vector machines
CN109993195B (en) * 2017-12-31 2024-04-12 国民技术股份有限公司 Side information processing method and device, terminal and computer readable storage medium
CN108446555A (en) * 2018-02-11 2018-08-24 复旦大学 The method that hardware Trojan horse is monitored in real time and is detected
CN108828325B (en) * 2018-04-23 2019-07-16 电子科技大学 Hardware Trojan horse detection method based on FPGA Clock Tree electromagnetic radiation field
CN109272502B (en) * 2018-09-28 2022-05-20 电子科技大学 PCB hardware safety detection method based on temperature field effect
CN109522755A (en) * 2018-10-09 2019-03-26 天津大学 Hardware Trojan horse detection method based on probabilistic neural network
CN109858246B (en) * 2018-12-24 2022-06-14 福州大学 Classification method for control signal type hardware trojans
CN110059504B (en) * 2019-03-01 2021-02-26 西安电子科技大学 Hardware Trojan horse detection method and device
CN113010883B (en) * 2019-12-20 2022-10-25 天津大学 Hardware Trojan horse detection method based on self-organizing neural network
CN117034374A (en) * 2023-08-28 2023-11-10 绍兴龙之盾网络信息安全有限公司 LM-BPNN hardware Trojan detection method and system based on PSO

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102662144A (en) * 2012-03-30 2012-09-12 北京大学 Activity measurement-based hardware trojan detection method
CN102799813A (en) * 2012-06-29 2012-11-28 武汉大学 Hardware Trojan horse detection system based on puf
CN102831349A (en) * 2012-08-23 2012-12-19 武汉大学 Characteristic value processing method for hardware Trojan detection

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9081991B2 (en) * 2011-03-23 2015-07-14 Polytechnic Institute Of New York University Ring oscillator based design-for-trust

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102662144A (en) * 2012-03-30 2012-09-12 北京大学 Activity measurement-based hardware trojan detection method
CN102799813A (en) * 2012-06-29 2012-11-28 武汉大学 Hardware Trojan horse detection system based on puf
CN102831349A (en) * 2012-08-23 2012-12-19 武汉大学 Characteristic value processing method for hardware Trojan detection

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
Exhibition(DATE)》.2012,第965-970页. *
Test in Europe Conference &amp *
Yier Jin et al..Post-Deployment Trust Evaluation in Wireless Cryptographic ICs.《Design,Automation &amp *
刘长龙等.基于侧信道分析的硬件木马建模与优化.《华中科技大学学报(自然科学报)》.2013,第41卷(第2期),第53-57页. *

Also Published As

Publication number Publication date
CN103198251A (en) 2013-07-10

Similar Documents

Publication Publication Date Title
CN103198251B (en) Hardware Trojan horse recognition method based on neural network
CN103150498B (en) Based on the hardware Trojan horse recognition method of single category support vector machines
Ellis et al. Techniques for improved heavy particle searches with jet substructure
CN101226590B (en) Method for recognizing human face
CN101976360B (en) Sparse characteristic face recognition method based on multilevel classification
Małek et al. The VIMOS Public Extragalactic Redshift Survey (VIPERS)-A support vector machine classification of galaxies, stars, and AGNs
CN105897517A (en) Network traffic abnormality detection method based on SVM (Support Vector Machine)
CN109858477A (en) The Raman spectrum analysis method of object is identified in complex environment with depth forest
CN103293141A (en) A liquor vintage recognition method based on a fusion technology of ion mobility spectrometry/ mass spectrometry/ Raman spectroscopy
CN104374738A (en) Qualitative analysis method for improving identification result on basis of near-infrared mode
CN104504412A (en) Method and system for extracting and identifying handwriting stroke features
CN105701470A (en) Analog circuit fault characteristic extraction method based on optimal wavelet packet decomposition
CN108846307B (en) Microseism and blasting event identification method based on waveform image
CN104374739A (en) Identification method for authenticity of varieties of seeds on basis of near-infrared quantitative analysis
Imani et al. Principal component discriminant analysis for feature extraction and classification of hyperspectral images
CN105116397A (en) Radar high-resolution range profile target recognition method based on MMFA model
CN109995501B (en) Side channel analysis method and device, terminal and computer readable storage medium
CN106992965A (en) A kind of Trojan detecting method based on network behavior
CN104376325A (en) Method for building near-infrared qualitative analysis model
CN106596513A (en) Tea leaf variety identification method based on laser induced breakdown spectroscopy
CN109446848A (en) A kind of hardware Trojan horse detection method based on Principal Component Analysis
CN109242010A (en) A kind of sparse study RCS sequence characteristic extracting method
CN106250913A (en) A kind of combining classifiers licence plate recognition method based on local canonical correlation analysis
CN104318224A (en) Face recognition method and monitoring equipment
CN103295007B (en) A kind of Feature Dimension Reduction optimization method for Chinese Character Recognition

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20150708

Termination date: 20160328

CF01 Termination of patent right due to non-payment of annual fee