Author: Authors: Fatemeh Marzani, Soheil Dehghani
Figure: Participants actively engaged at the WP 5 meeting.
The WP5 meeting brought together researchers and partners to share progress, capture partner interests, and align next steps for privacy-aware mobility analytics within XCARCITY.
This took place on the 29th of September 2025, in Apeldoorn at CVD. THe participants included: WP5 researchers and partners and Jennifer Faber (The Programme Officer from NWO)
Presentation Highlights
- Victor L. Knoop (with Dingshan Sun and Marco Rinaldi) presented a TRB paper on a scalable augmented link-based super-network for multi-modal traffic assignment. Path-dependent elements (transfers, parking, base fares) are modeled via augmented transfer links, avoiding explicit path enumeration. User equilibrium is posed as a variational inequality and solved with an extra-gradient algorithm. Case studies—including a ride-hailing mode—match path-based results while scaling better on complex, multi-mode networks.
- Soheil Dehghani (PhD candidate, University of Twente) presented Privacy-Preserving Mobility Monitoring Using Wireless Signals. The work explores using wireless data (e.g., Wi-Fi probe requests) to estimate footfall and crowd flow while protecting privacy. Key challenges include privacy risks, MAC address randomization, and higher-layer encryption. Next steps focus on analyzing features across wireless technologies to develop reliable, privacy-compliant, real-time mobility metrics for XCARCITY.
Fatemeh Marzani reported WP5 progress on privacy-preserving people counting and cross-camera flow estimation without explicit identification. The talk addressed: (1) safeguarding privacy, especially for small groups (e.g., random sampling); (2) quantifying privacy with differential-privacy parameters (ε, δ); and (3) selecting identifiers, favoring feature embeddings over raw images. A method for deriving stable, privacy-preserving identifiers enables intersection counting between locations to estimate footfall and flows. Next steps include a position paper, sensitivity studies (pose/lighting/quality), large-scale robustness testing, and adversarial-risk analysis (reconstruction, re-identification
Partners’ Interests
Partners discussed desired, transferable outcomes from each research stream—deployable metrics, integration interfaces, and compliance assurances—and identified practical collaboration opportunities across institutions.
Alignment of Research & Next Steps
Participants aligned on milestones, support needs (data access, evaluation protocols), and expectations for the next period. Actions include defining integration paths into the broader XCARCITY stack and establishing a lightweight privacy/compliance review prior to field deployments.