Creating privacy preserving heatmap with functional encryption - example

This repository demonstrates how the GoFE functional encryption (FE) library can be used for creating heatmap from location data of users in a way that the data of each individual is anonymous and encrypted. The server is able to perform computations on encrypted data together with keys given by all the users decrypting only the sum of all the locations, i.e. creating a heatmap.

Particularly, we simulate users of London Underground giving encrypted location information to the server allowing the server to create only the sum of all their locations while the individual's locations remain anonymous to the server as well as to the other users. With such data the server gains information about the usage of each station making it able to detect increased traffic, congestions, etc., while preserving the privacy of the users.

Input data

In this repository one can find data about the stations of London Underground (london_stations.csv) and train lines connecting them (london_connections.csv). The data is collected with Python3 script london_create_data.py, which additionally creates location information about users. The latter is created in a random way, each user randomly chooses a starting station and by a mix of shortest paths and random choices travels to a final station. For example, data of one user can be visualized in the following image showing the path traveled by one user:

To be more precise, the data of each user is a vector of zeros and ones, each input of the vector indicating if the user has traveled via the corresponding station. The data of all such vectors is saved in london_paths.txt. Note: this is a private information of every user which in a practical scenario would not be collected.

Running the example

1. Build the example by running:

$ go get github.com/fentec-project/FE-anonymous-heatmap

2. This will produce the FE-anonymous-heatmap executable in your $GOPATH/bin. If you have $GOPATH/bin in your PATH environment variable, you will be able to run the example by running command FE-anonymous-heatmap from the root of this repository.

Otherwise just call:

$ $GOPATH/bin/FE-anonymous-heatmap

The code will do the following:

• It will read the data from london_paths.txt and simulate the interaction between clients and the server. Each client has a vector describing his location data, which he does not wants to disclose. The goal of the server is to obtain the heatmap of all the locations.
• Clients first need to agree on a common shared secret. This is needed since there is no central server which can assign secret keys to the users; if such a server existed the scheme would not be anonymous any more. Creating a common secret is done by each user publishing a public key. Then each user can collect the public keys of all the others and creates a shared secret with his private key.
• Each user then encrypts his vector of locations with his secret.
• Each user creates a partial key which can be used only for the decryption of the sum of all the the location vectors.
• Finally, the server collects partial keys of all the users. With all the keys it is able to decrypt the sum of all location vectors giving it the data for the heatmap. No individual data can be decrypted with these keys.

Running the above script should output the following:

reading the data; numer of clients: 100
clients agreed on secret keys
simulating encryption of 100 clients
clients encrypted the data
clients created keys for decrypting heatmap
heatmap decrypted:
[1 2 2 7 1 2 1 3 4 5 5 4 4 13 2 3 5 1 0 12 3 4 1 4 4 0 5 4 9 3 5 3 4 1 8 2 5 2 3 0 6 4 3 5 6 1 2 0 12 4 7 1 6 1 3 6 3 8 4 2 2 1 1 2 4 2 2 3 1 2 4 5 4 1 6 1 2 3 1 3 3 11 1 11 2 7 1 1 0 0 2 2 5 1 2 5 5 2 3 0 2 2 5 2 0 3 3 1 2 1 1 3 6 5 1 0 1 5 1 3 2 6 7 8 1 0 7 2 0 5 2 1 0 5 0 3 4 10 0 1 1 16 3 6 3 4 3 3 6 3 13 2 2 2 0 3 7 12 1 8 0 0 6 5 6 1 6 10 6 1 4 2 6 1 2 2 4 10 5 3 4 11 2 4 2 1 6 3 3 10 2 5 0 1 5 1 15 2 4 12 7 1 2 2 6 2 9 2 5 5 7 3 4 1 9 7 5 1 2 1 0 1 5 4 4 5 3 3 4 5 0 4 1 1 1 1 0 4 4 5 4 2 1 0 2 4 1 1 3 3 9 6 5 0 13 3 1 3 3 1 2 3 2 7 4 4 9 3 4 11 9 5 4 1 0 6 2 2 3 7 3 22 5 3 1 3 1 2 4 4 2 5 3 1 2 1 1 4 15 4 3 4]

Explanation: the code simulates the above described process for 100 users. There are 302 London Underground stations included in this example, hence the output is a 302-dimensional vector describing how many users traveled through each station.

Visualization

The obtained heatmap vector can be visualized on the network of London Underground with colors indicating the traffic through each station. This can be achieved by running a provided Python3 script. First install the dependencies:

$ pip3 install networkx matplotlib

Navigate to the root of this repository and run the visualization script:

$ cd $GOPATH/src/github.com/fentec-project/FE-anonymous-heatmap
$ python3 london_visualize.py

The image is now saved in heatmap.png. It should look like:

Generating training data

If one wishes to re-create the random data, this can be cone by running the following Python3 script:

$ python3 london_create_data.py