From motion sensor data to mobility privacy: BIFOLD contributes a study on realistic tracking risks in public transport
The 28th International Conference on Modeling, Analysis and Simulation of Wireless and Mobile Systems (MSWiM 2026) will be held in Paris, France, from October 26 to 30, 2026. At the conference, Anne Josiane Kouam will present the paper “On the Feasibility of Reference-Free Tracking via Motion Sensors in Public Transport.” The paper assesses under realistic conditions whether motion sensor data from smartphones can infer the routes people travel by bus or train.
Paper:
Title: On the Feasibility of Reference-Free Tracking via Motion Sensors in Public Transport
Preprint: mlsec.tu-berlin.de/docs/2026-mswim.pdf
Authors: Anne Josiane Kouam (Inria & TU Berlin), Stefan Czybik (BIFOLD, TU Berlin), and Konrad Rieck (BIFOLD, TU Berlin).
Presentation: October 29, 2026, 13:30, Session 11: Performance Evaluation
Summary
Smartphone apps can often access accelerometer and gyroscope data without special permission. Earlier studies showed that such data can reveal which bus or train route a person has taken, but they relied on reference recordings of the transit network and a fixed phone position. The authors present an unsupervised pipeline that uses only publicly available transit information and structural features of the sensor signals, such as stops and curves, to narrow down and rank possible routes.
Tested on 172 real recordings across 21 lines in the public transport networks of Berlin and Stuttgart, the study shows that exact route reconstruction remains difficult under realistic conditions. Nevertheless, even coarse sensor data substantially reduce uncertainty about a journey, often leaving only a small set of plausible routes. The size of the risk depends largely on the structure of the transit network: smaller, less ambiguous networks increase the exposure of mobility data.