Article révisé par les pairs
Résumé : This paper presents a crowd monitoring systembased on the passive detection of probe requests. The systemmeets strict privacy requirements and is suited to monitoringevents or buildings with a least a few hundreds of attendees. Wepresent our counting process and an associated mathematicalmodel. From this model, we derive a concentration inequalitythat highlights the accuracy of our crowd count estimator. Then,we describe our system. We present and discuss our sensorhardware, our computing system architecture, and an efficientimplementation of our counting algorithm—as well as its spaceand time complexity. We also show how our system ensures theprivacy of people in the monitored area. Finally, we validateour system using nine weeks of data from a public libraryendowed with a camera-based counting system, which generatescounts against which we compare those of our counting system.This comparison empirically quantifies the accuracy of ourcounting system, thereby showing it to be suitable for monitoringpublic areas. Similarly, the concentration inequality provides atheoretical validation of the system.