Privacy-Preserving WiFi Fingerprint-Based People Counting for Crowd Management

The practice of people counting serves as an indispensable tool for meticulously monitoring crowd dynamics, enabling informed decision-making in critical situations, and optimizing the management of urban spaces, facilities, and services. Beyond its fundamental role in safety and security, tracking people’s flows has evolved into a necessity for diverse business applications and the effective administration of both outdoor and indoor urban environments. In the ongoing exploration of the study, emphasis is placed on employing a passive counting technique. This method leverages WiFi probe request messages emitted by smart devices to assess the number of devices, providing a reliable estimate of the number of people in a specific area. However, it is crucial to acknowledge the dynamic landscape of privacy regulations and the concerted efforts by leading smart-device manufacturers to fortify user privacy, as evidenced by the adoption of MAC address randomization. In response to these considerations, an enhanced iteration of the WiFi traffic generator has been introduced. This upgraded version is designed to generate realistic datasets with ground truth, aligning with the evolving privacy landscape. Additionally, leveraging a profound understanding of probe requests and the capabilities of the designed generator, a novel crowd monitoring solution that incorporates machine learning techniques, named ARGO, has been developed. This innovative approach effectively addresses challenges posed by randomized MAC addresses, incorporating Bloom filters to ensure a formal “deniability” that complies with stringent regulations, including the European GDPR. The proposed solution adeptly addresses the pivotal task of people counting by harnessing WiFi probe request messages. Significantly, it prioritizes users’ privacy, aligning with the foundational principles outlined in regulations such as the European GDPR.