gaoyang23nj / diffprivacyoppnet

Differential Privacy for OppNet

This repository provides the simulations about the routing algorithm in OppNet. It also supports the auxiliary information privacy-preserving method based the differential privacy.

The simulation is implemented in Python. It depends on the python packages, e.g., numpy, sklearn, pandas and cvxopt. The figuring code is implemented in Matlab.

Dataset

The dataset about the bicycle trips are published in this folder.

The bike stations are distributed in Pukou area (117 stations) and Qiaobei area (99 stations), Nanjing, China. The published dataset has been pre-processed. In each trip, the check-in station and the check-out station can be identified.

The number of trips between Pukou area and Qiaobei area is few (about 4,000 trips). However, the number of trips among the stations in Qiaobei area is more than 580,000 according to the file data_qiaobei.csv. In the other word, the trips in Pukou area and Qiaobei area are relatively independent.

Simulation

The main programs in this folder includes 'MainSimulator_XXX.py' in 'DiffPrivacyOppNet/01-code/Main/'. For example, conducting 'MainSimulator_varyttl_OpDP.py' will present the varying routing performace with the increasing TTL (the message time lifespan or the message time-to-live).

The verison of the ONE simulator, which supports the one-way connection (message transmission) is provided in here.

Figuring

The figuring programs in this folder is the '*.m' file, which should be conducted by Matlab.