Collecting Geospatial Data under Local Differential Privacy with Improving Frequency Estimation

Prerequisites

• Anaconda Python 3.8
• MATLAB R2020b
  • Install MATLAB Engine API for Python with referring to https://www.mathworks.com/help/matlab/matlab_external/install-the-matlab-engine-for-python.html
• Clone the project

  git clone https://github.com/daeyounghong/square-mechanism
  

Configuration

Create config.json in the project directory and write the contents as follows:

{
  "data_dir": "data",
  "eps": 12.5,
  "grid_shape": [600, 300]
}

• "data_dir": the path of a directory where dataset file and output files are stored.
• "eps": the privacy budget used for running the algorithm.
• "grid_shape": the grid shape used in the proposed postprocessing for frequency estimation based on convex optimization.

Preprocessing

• Download Gowalla dataset
• Place Gowalla data file loc-gowalla_totalCheckins.txt on data_dir of config.json
• Run the following command:

  python prepro.py
  

• The preprocessed file is stored in data_dir of config.json

Running the Algorithm

• Run the following command:

  python run_alg.py