Monitoring Large Crowds With WiFi: A Privacy-Preserving Approach

Abstract

This paper presents a crowd monitoring system based on the passive detection of probe requests. The system meets strict privacy requirements and is suited to monitoring events or buildings with a least a few hundreds of attendees. We present our counting process and an associated mathematical model. From this model, we derive a concentration inequality that highlights the accuracy of our crowd count estimator. Then, we describe our system. We present and discuss our sensor hardware, our computing system architecture, and an efficient implementation of our counting algorithm-as well as its space and time complexity. We also show how our system ensures the privacy of people in the monitored area. Finally, we validate our system using nine weeks of data from a public library endowed with a camera-based counting system, which generates counts against which we compare those of our counting system. This comparison empirically quantifies the accuracy of our counting system, thereby showing it to be suitable for monitoring public areas. Similarly, the concentration inequality provides a theoretical validation of the system.

Statistical Approach to Estimating Audience from MAC-Randomized WiFi Probe Requests

“…Pioneering studies of a detailed PR frame structure, e.g., [1][2][3], have enabled possible workaround solutions. One line of attack is to exploit existing privacy flaws to perform MAC de-randomization [4,5]; however, a consensus is emerging that data privacy is here to stay and that adopted solutions must be immune to randomization [6][7][8][9][10].…”

Tools for Ground-Truth-Free Passive Client Density Mapping in MAC-Randomized Outdoor WiFi Networks

“…Again, as introduced in Section 1.4, it has recently been demonstrated theoretically in [1,6] and observed empirically in [1,6,8] that due to the statistical independence of the client density and client device emission probability distributions, a simple proportionality exists, in the limit of a large number of clients, between the number of randomized MAC clients, C, and the corresponding number of randomized MAC PR, P, that is, C = P/X, where the proportionality constant X is understood to be the mean device PR emission probability for the time window concerned. This observation is currently the key to understanding client numbers and client densities in the era of data privacy.…”

“…A consensus is growing that data privacy is here to stay and that researchers would do better to concentrate on intrinsically non-device-specific privacy-preserving approaches. Indeed, this is the philosophy adopted in the present work as well as in other recent contributions [1,[6][7][8].…”

“…It is found that for sufficiently large client populations, the number of PRs is linearly proportional to the true number of clients. The problem remaining then amounts to determining the appropriate proportionality constant, which can be achieved either by comparison to a ground truth obtained from other sensors, such as cameras, entry monitors, etc., as in [6][7][8], or without ground truth, albeit with reduced accuracy, by exploiting observed periodicities in plots of PR counts versus time [1]. This type of approach forms the starting point for the mobile client localization techniques that will be developed in the present work.…”

“…On the one hand, the appropriate value of X for a particular site can be obtained using ground truth from alternate counting systems, such as cameras. This was the technique employed in [6,8] for indoor localization in networks having relatively high AP density and equipped with complementary ground truth sensors located in the various zones covered by the WiFi network.…”