PrivCount is an independent implementation of the PrivEx Secret Sharing (S2) variant, that has been customized in order to be able to aggregate a large variety of statistical counts from Tor while providing differential privacy guarantees. For more information, see the associated publication:
Safely Measuring Tor
23rd ACM Conference on Computer and Communication Security (CCS 2016)
Rob Jansen and Aaron Johnson
For more information about PrivEx, see:
PrivEx: Private Collection of Traffic Statistics for Anonymous Communication Networks
21st ACM Conference on Computer and Communications Security (CCS 2014)
Tariq Elahi, George Danezis, and Ian Goldberg
Base components:
system libs: libssl libssl-dev cffi
python libs: pyyaml, twisted, pyopenssl, service-identity, cryptography
Optional graphing extensions (required only for the plot subcommand):
system libs: libpng libpng-devel, #TODO this list is incomplete
python libs: numpy, matplotlib
System libs can be install with apt-get, yum, brew, etc. Python libs can be installed with pip,
as we explain below.
I recommend using virtual environments to isolate the python environment and avoid conflicts. Run the following from the base directory of this package (i.e., the same location of this README). (If you're using OS X or ubuntu, you might need to use "sudo -H" before some of the pip commands.)
pip install virtualenv
virtualenv --no-site-packages venv
source venv/bin/activate
pip install -r requirements.txt
# if you want to use the optional privcount plot command
pip install -r requirements-plot.txt
pip install -I .
test/run_test.sh . # run the unit tests to check your setup
deactivate
To run PrivCount, simply activate the virtual environment that you created earlier and then run PrivCount as normal. For example:
source venv/bin/activate # enter the virtual environment
privcount --help
...
deactivate # exit the virtual environment
See test/README.markdown for notes about testing PrivCount in a private local deployment, or just run the unit tests:
source venv/bin/activate # enter the virtual environment
test/run_test.sh -I .
deactivate # exit the virtual environment
The notes below are outdated and will no longer work!
Example of the global section for a privcount-config.yml file, which all nodes need:
global:
start_time: 1452520800 # 2016-01-11 at 2pm UTC
epoch: 604800 # (1 week = 604800 seconds) the safe time frame of stats collection for all stats
clock_skew: 300 # seconds - to deal with clock skews and latency
Generate key and create self signed cert in a new base directory:
mkdir privcount_ts
cd privcount_ts
touch privcount-config.yml # add above global config in here
openssl genrsa -out ts.key 4096
openssl req -new -x509 -key ts.key -out ts.cert -days 1825
Choose an address W1.X1.Y1.Z1 and port P1 that is accessible on the Internet, and append the
following as a new section under the global section of the privcount-config.yml file:
tally_server:
listen_port: P1 # open port on which to listen for remote connections from TKSes
key: 'ts.key' # path to the key file
cert: 'ts.cert' # path to the certificate file
results: 'results.txt'
Then run PrivCount in tally server mode:
privcount privcount-config.yml ts
Generate key and create self signed cert in a new base directory:
mkdir privcount_tks
cd privcount_tks
touch privcount-config.yml # add above global config in here
openssl genrsa -out tks.key 4096
openssl req -new -x509 -key tks.key -out tks.cert -days 1825
Choose an address W2.X2.Y2.Z2 port P2 that is accessible on the Internet, and append the
following as a new section under the global section of the privcount-config.yml file:
tally_key_server:
listen_port: P2 # open port on which to listen for remote connections from DCs
key: 'tks.key' # path to the key file
cert: 'tks.cert' # path to the certificate file
tally_server_info: # where the tally server is located
ip: W1.X1.Y1.Z1
port: P1
Then run PrivCount in tally key server mode:
privcount privcount-config.yml tks
Create a new base directory:
mkdir privcount_dc
cd privcount_dc
touch privcount-config.yml # add above global config in here
Choose a local port L that will listen for connections from Tor.
data_collector: listen_port: L # local port on which to listen for local connections from Tor noise_weight: 1.0 # distribute noise among all machines / data collectors tally_server_info: # where the tally server is located ip: W1.X1.Y1.Z1 port: P1 tally_key_server_infos: # where the tally key servers are located - ip: W2.X2.Y2.Z2 port: P2 statistics: ... # see test/privcount-test-config.yaml