Large-scale Wi-Fi and Bluetooth data collection for reconstructing passenger flows

ABSTRACT

For a public transport operator, estimating passenger flows is vital to plan inclusive and efficient service. Understanding where and when to optimally allocate resources results in better customer experiences, further shaping and optimizing public transport services. To this end, the project Mobile Data Fusion aims at developing a technique to collect, process, and merge data from different sources. The data are used to inform public transport operators about passenger demand, by combining Wi-Fi and Bluetooth signals from customer devices with time-of-flight counting sensors. The project had access to a large study area in Kassel, Germany, with a fleet of 50 vehicles servicing eight passenger lines, all equipped with automatic passenger counting sensors. In the course of this study, we found out that Bluetooth signals did not offer any significant insight , while Wi-Fi signals could be used to reconstruct a trend, albeit not the absolute occupancy level. In this paper, we present our hardware setup and a subset of our data set to show how the Wi-Fi and Bluetooth signals correlate with the measured occupancy. The ongoing data collection will provide additional material for future studies, and answer some of the pending questions discussed in this work.

KEYWORDS:

• Public transport
• Wi-Fi sniffing
• Bluetooth sniffing
• passenger count
• passenger flows

Acknowledgments

This research has been funded by the Federal Ministry of Transport and Digital Infrastructure of Germany (BMVI) within the mFund funding line under grant number #19F2081C. The authors acknowledge NVV Kassel for enabling the data collection and providing access to the vehicle fleet and INIT Group for supporting the publication of the results of the field test. The authors would also like to thank Pascal Faller, Sebastian Strauß, Larissa Hauer, and Henriette Kissling, who contributed to the initial conception of the project and worked on its back-end infrastructure, and all their colleagues in the INIT Research&Technology team for issuing suggestions and for proof-reading this manuscript. Additionally, the authors would like to thank Lucia Pintor, Luigi Atzori (Università di Cagliari) and Marco Uras (WiData SRL) for productive discussions about the topic of MAC address randomisation and possible technical workarounds. The authors acknowledge OpenStreetMap for the background map data used in Figures 1 and 5: Map data copyrighted OpenStreetMap contributors and available from https://www.openstreetmap.org (OpenStreetMap contributors 2017).

Disclosure statement

All authors are, at the moment of manuscript submission, employed by INIT GmbH, Käppelestr. 4–10, 76131 Karlsruhe, Germany, which partly financed the project this work reports on and which produces some of the hardware components used in this project (board computer, passenger counting sensors).

Author contributions

The authors confirm their contribution to the paper as follows: study conception and design: J. Wendel, M. Quinting; data collection: F. Elgner, A. Demetrio, H. Hameister, D. Warzok; analysis and interpretation of results: A. Demetrio, H. Hameister, D. Warzok, J. Wendel; draft manuscript preparation: A. Demetrio. All authors reviewed the results and approved the final version of the manuscript.

Notes

1. www.blic.de/en/ (accessed 21 Oct 2022)

2. personal communication with the L. Pintor, one of the authors of the paper in question.

Additional information

Funding

This work was supported by the Federal Ministry of Transport and Digital Infrastructure of Germany (BMVI).