MAster assignment
Overcoming Wi-Fi MAC Randomization for Wi-Fi Based Crowd Flow Analytics
TYPE : MASTER t.b.a.
Period: Start date: as soon as possible
Student: Unassigned
If you are interested please contact:
Description:
Background:
Monitoring crowds of people using Wi-Fi signals transmitted by their smartphones has become widespread [1,2]. Probe requests broadcasted by Wi-Fi interfaces of smartphones are captured by sensing infrastructures and later on used for counting the crowds present in certain locations, as well as crowd flows developing between different locations [3]. Such monitoring used to be conveniently achievable, as probe requests contained MAC addresses of senders, allowing thus the accurate identification and counting of devices. In recent years, to address privacy concerns of users, many smartphone manufacturers introduced Wi-Fi MAC randomization [4,5], a technique which replaces true MAC addresses with random ones when broadcasting probe requests. For crowd-monitoring, this measure implies that devices can no longer be uniquely identified through their MAC address as they randomly choose and also change their addresses using a heterogeneous range of mechanisms.
A new generation of methods for counting crowds emerged, aiming to cope with randomized MAC addresses. These methods mostly rely on information contained in probe requests which can be used to group together signals sent by the same device. For example, in [6] authors discovered that there are certain fields which remain unchanged, so-called tagged parameters, despite the MAC address being randomized; at the same time, they exploit frame sequence numbers, which have an unchanged increasing pattern. In a similar attempt [7], ML techniques are used around the same invariants of probe requests. However, most of these unicity traits have been patched by manufacturers, increasingly making the aforementioned methods impractical. A promising recent work by Torkamandi et al. [8] implements Wi-Fi fingerprinting of devices based on the timing information of probe requests. The authors rely on the time between probe request bursts containing identical MAC address as a feature for clustering together different randomized MAC addresses corresponding to the same device.
Assignment:
In this project, taking into account the results of [8], several paths are to be investigated. First of all, [8] considers only inter-burst timings and MAC addresses for clustering. We would like to investigate the influence of also considering intra-burst characteristics, as well as other Wi-Fi interface behavioural information that may be available. Additionally, [8] considers only static, isolated environments, its usability being limited to footfall counting. We intend to explore how such an approach could work for counting crowd flows between different locations too, implying finding ways to accurately match clusters built at different locations.
References:
[1] Weppner, J., Bischke, B., & Lukowicz, P. (2016, September). Monitoring crowd condition in public spaces by tracking mobile consumer devices with wifi interface. In Proceedings of the 2016 ACM International Joint Conference on Pervasive and Ubiquitous Computing: Adjunct (pp. 1363-1371).
[2] Draghici, A., & Steen, M. V. (2018). A survey of techniques for automatically sensing the behavior of a crowd. ACM Computing Surveys (CSUR), 51(1), 1-40.
[3] Schauer, L., Werner, M., & Marcus, P. (2014, December). Estimating crowd densities and pedestrian flows using wi-fi and bluetooth. In Proceedings of the 11th International Conference on Mobile and Ubiquitous Systems: Computing, Networking and Services (pp. 171-177).
[4] Martin, J., Mayberry, T., Donahue, C., Foppe, L., Brown, L., Riggins, C., ... & Brown, D. (2017). A Study of MAC Address Randomization in Mobile Devices and When it Fails. Proceedings on Privacy Enhancing Technologies, 4, 268-286.
[5] Fenske, E., Brown, D., Martin, J., Mayberry, T., Ryan, P., & Rye, E. (2021). Three Years Later: A Study of MAC Address Randomization in Mobile Devices and when it Succeeds. Proceedings on Privacy Enhancing Technologies, 2021(3), 164-181.
[6] Nitti, M., Pinna, F., Pintor, L., Pilloni, V., & Barabino, B. (2020). iabacus: A wi-fi-based automatic bus passenger counting system. Energies, 13(6), 1446.
[7] Uras, M., Cossu, R., Ferrara, E., Bagdasar, O., Liotta, A., & Atzori, L. (2020, September). Wifi probes sniffing: an artificial intelligence based approach for mac addresses de-randomization. In 2020 IEEE 25th International Workshop on Computer Aided Modeling and Design of Communication Links and Networks (CAMAD) (pp. 1-6). IEEE.
[8] Torkamandi, P., Kärkkäinen, L., & Ott, J. (2021, March). An online method for estimating the wireless device count via privacy-preserving wi-fi fingerprinting. In International Conference on Passive and Active Network Measurement (pp. 406-423). Springer, Cham.