Graph-theoretic

Chinese README

1.Project description

This project deals with hardware Trojans in untrusted 3PIPs. Please refer our paper "Toward Building and Optimizing Trustworthy Systems Using Untrusted Components: A Graph-Theoretic Perspective" for more detail.

2. Description of content

2.1 Directory description

The complete contents of the directory are as follows:

├── data
│   ├── graphColor
│   │   ├── 0_0_cipr.txt
│   │   ├── 0_0_cip.txt
│   │   ├── 0_0_weight.txt
│   │   ├── ...
│   └── rawData
│       ├── 0.txt
│       ├── ...
│       ├── area.txt
│       └── weight.csv
├── graphAlg
│   ├── a.out
│   ├── graphColor.ipynb
│   ├── graph.h
│   ├── graph.cpp
│   └── util.h
├── output
│   ├── 1_0_res.txt
│   ├── ...
└── README.md

The above directories can be divided into:

• Dataset directory

  • rawData

  For the original data set catalog, weight.csvdifferent types of IP core data sets are produced for different manufacturers, and area.txtthe areas corresponding to different types of nodes.

  • graphColor

  For the graphColor.ipynboutput path

• Code directory

  • graphColor.ipynb

    Is a jupyter-notebook file, the main functions are as follows:

    • Data preprocessing and graph coloring
    • Relationship between color and node type
    • The relationship between colors and nodes
    • Weight matrix matching minimum weight in bipartite graph

  • graph.cpp

    It is the C ++ main program. Its specific functions are implemented in graph.hand util.h. The main functions are as follows:

    • Read data/graphColorthe data set in the directory
    • Output the following experimental results to the output directory
    • Optimize match to complete the optimal match based on bipartite graph
    • Minimize cycle and complete time-based IP core scheduling
    • Optimize area and complete area-based IP core scheduling

• Results directory

  • outputOutput results for experiment

2.2 Data description

• rawData

  Description of the original data set, which contains some benchmark data sets.

• graphColor

  graphColor.ipynbLoading each raw dataset will output 3 files:

  1_0_cip.txt, 0_0_cipr.txt,0_0_weight.txt

  1 represents the original dataset number, 0 represents the number of trusted IP cores, cip represents the color and node type dataset, cipr represents the color and node dataset, and weight represents the output weight dataset.

  Among them, cip and cipr are understood by columns, which are the nodes or node types corresponding to each color.

• output

  1_0_res.txtRepresents the experimental results of the first data set with 0 trusted IP cores

3. Instructions for use

3.1 Environmental requirements

• Python 3.x
  • jupyter-notebook
  • numpy
  • networkx
  • matplotlib
  • pandas
  • copy
• C ++
  • Compiler supporting at least C ++ 11

Our development environment is:

• Ubuntu18.04 + Anaconda3 + GCC8.3

3.2 How to use

• Please make sure the 3.1 environment is correct

• Please run the graphColor.ipynbfile and the experiment results will be stored under the graphColor folder.

• Please rungraph.cpp

  • It works as follows:

  g++ graph.cpp -std=c++11
  ./a.out 1 0
  

  1 represents the data set number, 0 represents the number of trusted IP cores.

• View experimental results (enter the output directory)