See Browse projects

Privacy-preserving crowd monitoring A WiFi-tracking project

Living Smart Campus

Project information

The project

We examine how to safely and responsibly collect WiFi tracking data such that it is virtually impossible to track an individual. In this way, we want to prevent the needless and privacy invasive WiFi tracking of smartphones as seems to increasingly become common practice in many cities.

The work is primarily done at the UT, in collaboration with Enschede and the University Polytehnica Bucharest (UPB), as well as with Bluemark Innovations.

We anticipate several solutions for safely tracking WiFi-enabled devices, but also solutions for gathering data sets of scanned devices from which it is virtually impossible to find out who the owner of a device is. To the end, we need to address several hard problems, including storing data such that even the storage servers may be compromised without harming the preservation of privacy, as well as securing collected MAC addresses in a privacy preserving manner. It is unclear as yet whether we will be able to succeed.

how to collaborate

There are several ways to collaborate, mostly by helping out on data sets: securing them and other hand trying to break our security solutions.

Want to participate in this project?
Research and analysis

My project updates

5 Apr 2018

Scaling up

It's been quiet for a while (too quiet), but we're now at the verge of setting up a system that will allow us to obtain tracking information from many, many access points spread across the campus. The big challenge is to collect data in such a way that we not only abide by Dutch and European laws, but that we can produce a safe and secure database from which it is (virtually) impossible to trace data back to an individual. This is still going to take a lot of research, but progress is being made.

Our first step is that we produce a database that contains securely encrypted network addresses. That database is accessible to only a small group of researchers that are working on privacy preservation. We are looking into developing techniques that can truly hide the whereabouts of individuals. One example is deliberately contaminating data such that aggregated statistics remain the same in comparison to the clean data, yet the contamination is such that inspecting individual data items is senseless. Another one that we're investigating is using pseudonyms that are so short that you'll see a lot of collisions, yet again the statistics are enough to draw conclusions. As a last example, we are also considering the generation of answers to a huge set of questions (and those are the only answers you'll get), but deriving information on individuals from those answers is provably (computationally) impossible.

A very important note: this research is done in close collaboration with LISA (Jeroen van Ingen), and as an exercise in complying with GDPR, we're going through the whole song-and-dance of ensuring that we're doing the right thing. 

Again, stay tune, as we're finally making good progress! Technical details will follow soon.

11 Dec 2017

WiFi scanners up-and-running

After a lot of efforts to see how we could get WiFi scanners working, we finally managed to solve two important issues:

  • Getting batteries attached to the scanners that are powerful enough, and that are charged while there is current to feed the lamp post to which the scanner is connected.
  • Get the scanners to talk over eduroam in order easily collect data.

The goal of this research is to make sure that we can ask interesting questions, notably about the flow of pedestrians across the campus, while completely preserving privacy. The details of our initial method will be explained soon, once we have run the first privacy tests.

6 Apr 2017

Outdoor scanners being tested

We finally, after a very long time, managed to get the equipment for installing outdoor scanners. The problem was that outdoor scanners need to be placed at convenient places, meaning also that they need to be powered. Lamp posts are ideal, but when the lights are switched off, so will our scanners. The solution is using powerful batteries that are (fully) charged during the night. We are now running tests to see if we can have a scanner operate 24/7, partly powered through a 20 Ah battery. Once successful, we'll start installing some 70 scanners around the campus.

Stay tuned.